Saturday, January 21, 2012

GISS: how to defend a 2.3 °C climate sensitivity

NASA's GISS has completed their 2011 global temperature dataset. According to GISS, 2011 was the 9th warmest year. The average warming trend in 1880-2011 was 0.60 °C per century; in the satellite era 1979-2011, it was 1.58 °C per century, not too much greater than those 0.14 °C per century according to the satellite teams.

GISS is an anomalous source because it considers 2010 to be the warmest year while 1998 – the warmest year according to pretty much everyone else – is 4th according to GISS. That's quite a difference. Here is their hitparade, as I calculated it with Mathematica:
1. {2010, 0.63083333},
2. {2005, 0.61916667},
3. {2007, 0.58416667},
4. {1998, 0.58},
5. {2009, 0.5675},
6. {2002, 0.56333333},
7. {2003, 0.55583333},
8. {2006, 0.55333333},
9. {2011, 0.51416667}
10. {2001, 0.48333333},
11. {2004, 0.4825},
12. {2008, 0.44},
13. {1997, 0.41},
14. {1995, 0.39083333},
15. {1990, 0.36},
16. {1991, 0.35083333},
17. {2000, 0.35},
18. {1988, 0.33},
19. {1999, 0.32833333},
20. {1996, 0.29583333},
21. {1987, 0.2775},
22. {1981, 0.265},
23. {1983, 0.26166667},
24. {1994, 0.23583333},
25. {1989, 0.21166667},
26. {1980, 0.19916667},
27. {1944, 0.18583333},
28. {1973, 0.13916667},
29. {1993, 0.13583333},
30. {1977, 0.13416667},
31. {1986, 0.1275},
32. {1992, 0.125},
33. {1953, 0.098333333},
34. {1979, 0.0975},
35. {1938, 0.096666667}
Apologies: I forgot to round the figures and I kept the braces not to waste too much time with formatting.

But there's some new manipulation I want to share with you: an argument for a "medium" value of the climate sensitivity, namely 2.3 °C per doubling, which is a mean value reached by a calculation in Trenberth et al. 2010 as well as Schmittner et al. 2011. It's in between the mean IPCC climate sensitivity and the climate sensitivity believed to be the most accurate one by skeptics like myself. It's pretty high.

Imagine that you look for the best fit that links the observed GISS temperatures in the month $$x=1\dots 1584$$ corresponding to the 1880-2011 period to the CO2 concentrations given by my formula (units of ppmv are omitted)
$c = 280 + 22.3 \exp\left(\frac{{\rm year}-1920}{57}\right)$ You get this:
$T = -0.415 °{\rm C} + 1.972 °{\rm C}\times\log_2 \left(\frac{c}{280}\right)$ The coefficient 1.972 °C in front of the base-two logarithm is the estimated climate sensitivity associated with the doubling. The value of 1.97 °C per doubling directly extracted from GISS is high – the largest one you can get from similar operations – but it's still lower than the IPCC lower bound of 2 °C.

But you could try a more contrived fit. Imagine that you allow a term proportional to the logarithm of the concentration but you also add another term linear in time – corresponding to some sources of temperature change that are very slow and that kept pretty much a constant rate for more than a century, maybe from the Little Ice Age. The idea is that the non-CO2-related faster processes average out so only the non-CO2-related (mostly natural) very slow (and thus linear) processes are important.

Then you get a more complicated interpolation:
\begin{align} T &= -0.411 °{\rm C} - 0.0011\times \Delta{\rm year} °{\rm C} + \\ &+2.302 °{\rm C}\times\log_2 \left(\frac{c}{280}\right)\end{align} Here, $$\Delta{\rm year}={\rm year}-1880$$. This extended formula may a priori be more sensitive to the expected "acceleration" caused by the higher growth of CO2 in recent decades. Note that we obtain a very small extra linear term that corresponds to a cooling trend by 0.11 °C per century. It's very small. Because the non-CO2-related terms are decreasing, the predicted CO2 sensitivity is raised to 2.3 °C per doubling to provide you with a compensation. The smallness of the extra linear term, regardless of the sign, may also be interpreted as a sign of "good correlation" of the temperature with the CO2 concentrations.

Let me admit that I think that this simple fit is the most robust empirically rooted argument in favor of a sensitivity above 2 °C that I have ever seen. All other trustworthy calculations, including the calculation without the extra linear term, yield values of climate sensitivity that is below 2 °C. All calculations yielding results above 2 °C were flagrantly biased, skewed, or they were not calculations at all. Could the climate sensitivity be around 2 °C?

Of course, as far as I can say, it could. It would probably mean that we will continue to see some underlying warming trend by 1.5 °C per century we are seeing today, reaching the "total warming by one sensitivity" a decade or two before 2100. At that point, the correlation of temperatures would probably become self-evident. Also, people would already know that a degree of warming isn't something to be afraid of.

Germany abolishing solar subsidies

Bloomberg and Business Week are among those offering the news that Germany will gradually decrease and by 2017 completely terminate subsidies for photovoltaic panels. The prices of solar companies' stocks all over the world dropped after the decision was publicized. Because Germany is also in the process of banning nuclear power plants, it seems that King Coal is returning to Deutschland as an Emperor Coal.