Watts Up With That (WUWT) has a recent post called Monckton asks IPCC for correction to AR4. This post relates to a letter sent to the IPCC by Christopher Monckton in which he asks them to correct an interpretation of a figure included in the IPCC’s 4th Assessment Report. The figure is below and it has been used to suggest that global warming has been accelerating. Christopher Monckton seems to suggest that the interpretation is wrong as one could get the same result if temperature anomalies were simply varying sinusoidally. The idea is that the if the latter part of your graph is during an upward phase in the sinusoid, it will show acceleration but wouldn’t necessarily imply accelerating global warming.
The first thing I would say, as I’ve said many times before, is that basing your assessment of global warming on temperature anomaly data alone is simplistic. Ocean heating and the melting of polar ice are also indicators of global warming. Also, even if it is sinusoidal, the figure above seems to show that we’re in an accelerating phase. If I move off in my car after stopping at a traffic light I will be accelerating. At some point in the future I will presumably decelerate and stop. That does not mean that I’m not accelerating at this moment in time.
I will say, however, that I do wonder why the IPCC chose to produce the graph in the manner that they did. By choosing decreasing time intervals, the error for the later time intervals will be much larger than the error in the longer time intervals that represent the earlier periods. It seems more sensible to stick with the same start point and simply change the end point. I thought I would do this using the Skeptical Science Trend Calculator. I considered both GISSTEMP data and HADCRUT4 data and considered the time intervals 1880-1950, 1880-1970, 1880-1990, and 1880-2013. The resulting tables are below. The values in the table are the mean trend and the minimum and maximum trends based on the 2σ errors (i.e., there is a 95% chance that the actual trend is between the minimum and maximum values).
So, what do we see in these tables? The trends for the first 3 times intervals are similar, for both data sets. There is some suggestion in the GISSTEMP data that the 1880-1990 trend is larger than in the two earlier intervals, but there is a large overlap. What is clear is that the 1880-2013 trend is larger (with 95% confidence) than the trend in all three earlier intervals (i.e., in both the GISSTEMP and HADCRUT4 data, the minimum trend for 1880-2013 is larger than the maximum trend for all 3 earlier time periods). I would add that I’ve always wondered why this analysis isn’t used to argue against the supposed “pause” in global warming. If, according to temperature anomaly data, global warming has paused since the mid-90s, how can the trend for 1880-2013 be larger than the trend for 1880-1990?
Anyway, it seems clear that the temperature anomaly growth has accelerated in the last 20 – 30 years compared to previous time intervals. It also seems that there is no real evidence for any large sinusoidal variation. Certainly, there doesn’t seem to be a time interval for which the trend is zero or negative, which is what you would require if you want to suggest that it is all some multi-decade oscillation. So, as far as I can tell the IPCC interpretation is fine. It’s clear that the rate at which the temperature anomaly changed with time increased in the last few decades compared to earlier time periods. That, in my book, is called acceleration.
Addendum – 6 May 2013
One of comments below asks if I could present this graphically. I had a problem yesterday as my code was giving a different trend to that of the Skeptical Science Trend Calculator (SSTC). I re-downloaded the GISSTEMP data and re-calculated the trends and they now match that of the SSTC. What I have below is a figure showing the monthly GISSTEMP temperature anomaly data (dots), together with the 4 trend lines (1880-1950, 1880-1970, 1880-1990, 1880-2013). You can’t see the 1880-1950 one as it is on the 1880-1970 line. There is a small, but not statistically significant, increase in the trend for 1880-1990, and then quite a substantial increase for 1880-2013. As shown in the table above, even including the 2σ errors, the 1880-2013 trend does not overlap with any of the earlier trends. That, to me, would be an indicator that global warming has been accelerating in recent years.