April 2008 Update on Global Temperature, Part 4
Posted by The Diatribe Guy on April 22, 2008
In part 3 I showed the different trend lines associated with the 60-month raw anomaly data, the most recent peak (or trough) 60-month trend line, and the trend of the 60-month slopes in the most recent trend from the time period just before the most recent peak or trough.
I have done the same thing with 120-month trends, 180-month trends, 240-month trends, 300-month trends, and 360-month trends. Now, I will state up front that, in general, all of these graphs look similar. However, I am going to present them because the trend lines differ depoending on the period used. This helps demonstrate how any broad speculations of future trend lines based on any singluar set of data is clearly questionable, as any such trend line depends on the starting point of the data. However, it is relevant also to show that in each of the measures – to one degree or another – the latest peak/trough in the trend line is indeed a peak. So, the current X-month trend we are seeing is lower, to some extent, than the most recent peak value. The trend of the slope line will demonstrate the speed with which the slope is changing.
The other reason I will present them all is because these trends and slopes become an underlying basis for my methodology in predicting future temperature anomalies and trends. Simply projecting out a trend line is clearly an inferior method of projecting anomalies. History shows us that the trends lines are continually changing. I have incorporated past value changes into a predictive model, which will be explained and demonstrated at a future time.
For now, here are the visuals, for your viewing pleasure:
The current 120-month raw anomaly slope of 0.1379 compares to the peak of 0.3346 in May 2002. Since Early 2002, the slope has changed on the successive 120-month trend lines. The slope of that trend line is -0.0024. See previous parts of the series for a discussion of the units on the trend values.
The current 180-month trend has a slope of 0.2027, down from a peak of 0.2602 last March 2007. The trend here is on a shorter period and probably less meaningful, but still, the slope of the changing trend line is -0.0047.
The current trend of 0.1595 is down from the most recent peak value of 0.1755 in March 2004. Since then the trend down in the slope values is -0.0002. As the trend line periods lengthen, incremental changes to the trends become much lower, and so it is to be expected that, even in a declining trend, it will take a longer period of time for substantial movement to take place in the slope of the trend line. So, even though the value change is less, it is nonetheless significant. There are some interesting insights into this once we start looking at the predictive model.
The current 300-month slope is 0.1584, compared to the most recent peak of 0.1634 in the period ending February 2007. Again, not a long trend period to view, but the slope is -0.0005. The r-squared value of over 0.88 is a very good fit, but we can see that the latest point is quite a disruption, due to the unexpectedly high March 2008 anomaly.
The 360-month trend line hasn’t changed significantly, largely because of the very low anomalies in the beginning of the period, along with the high anomalies in recent years. It will take a little longer for the colder anomalies to drop off in the early period, but in a couple years, should anomalies drop, we could see a more rapid and significant change in the 30-year trend line. For now, we observe that the current slope is 0.1417, down from its recent peak value of 0.1503 in the period ending June 2005. The trend in the slope change is -0.0002, but the R-squared is lower here due to some fluctuations that took place. Still, the evaluation of the 30-year trends yields predicted anomalies in line with other trend periods, and seems to have predictive value when utilized properly.
So there you have it. Enjoy the charts, and if there is any clarification needed, just ask. In the next installment I will delve into the model I have put together and provide the predicted future anomalies, and the impact I expect this to have on the future trend lines.