## Wednesday, January 18, 2017 ... /////

### GISS: 1998-2016 comparison suggests a trend of 2 °C per century

Thursday update: British HadCRUT4 have completed their 2016 data, too. The last column contains the annual averages. The difference from GISS is significant. 2016 was only 0.013 °C (GISS: 0.13 °C!) warmer than 2015. December 2016 was 0.432 °C (GISS: 0.30 °C) cooler than December 2015. And 2016 was 0.237 °C (GISS: 0.36 °C) warmer than 1998, indicating just 1.3 °C (GISS: 2 °C, satellites: 0.11 °C) of warming per century!
While Czechia is enjoying the best skiing season – when it comes to the snow conditions – in years (Ore Mountains and the Bohemian Forest often provide skiers with up to 150 cm of snow) and I've exploited this fact as well, The New York Times told us about a press conferences by NOAA and NASA today that finally announced the temperature data for 2016.

GISS temperature anomalies, 1880-2016, in multiples of 0.01 °C

On January 3rd, I mentioned that both satellite-based teams quantifying the global mean temperature (UAH AMSU, RSS AMSU) concluded that 2016 was 0.02 °C warmer than 1998. These were otherwise very similar "end of a strong El Niño years" separated by 18 years. According to these numbers and nothing else, one could estimate that the warming per century is some 0.11 °C, a negligible amount.

The GISS data derived from surface measurements (weather stations for the land and some other gadgets in the ocean) ended up with a very different number than 0.02 °C for the difference between the temperatures in 2016 and 1998.

The GISS temperature anomaly for 2016 was 0.99 °C, about 0.13 °C warmer than the anomaly in 2015. It's a significant increase but the increase between 1997 and 1998 was larger still, 0.15 °C. According to GISS, the temperature in 2016 was 0.36 °C warmer than in 1998. You may see that the ratio is almost exactly 0.02 °C per year or 2 °C per century. This deduced trend is almost 20 times larger than that from the satellites but even if we trusted the linear extrapolation, it would still be far from a tragedy if the temperature in 2116 were 2 °C warmer than in 2016.

While the El Niño year 2016 was much warmer according to the terrestrial measurements, the post-El-Niño cooling seems rapid according to these measurements, too. As far as the temperature anomalies go, December 2016 with its 0.81 °C anomaly was a whopping 0.30 °C cooler than December 2015 and a Burger King Double Whopper 0.54 °C anomaly-cooler than February 2016, the highest-anomaly month on the GISS record.

The growing discrepancy between the satellites and the terrestrial measurements could be naively interpreted as bad news for science – there has to be some big mistake at least in one class of these measurements. But this conclusion isn't really right. Their different stories about the global mean temperature probably aren't due to a "mistake" but due to their different definition of the global mean temperature. The terrestrially measured temperature is increasing at this rate approximately 2 °C per century while the rate is less than 1.5 °C – and, according some proxies, even much smaller than that – according to the satellites. That's not a contradiction because they mean different quantities by the "global mean temperature".

There is no "canonical" global mean temperature. It's an artificial quantity whose detailed value – and whose detailed change in 18 or 100 years – significantly depends on all the details about how the global mean temperature is defined and measured. I've mentioned that 2016 was 0.02 °C warmer than 1998 by satellites but by 0.36 °C warmer than 1998 according to the terrestrial measurements. You may pretty much say that the difference of the annual global mean temperatures in 2016 and 1998 was 0.19 °C plus minus 0.17 °C. It's still pretty much compatible with the zero at the one-sigma level.

In Czechia, not only prisoners had to be exploited to remove the snow. In Jablonec, the famous "Stalin's hands" (a Soviet-made vehicle/elevator on the picture) had to be put to business, too.

Bonus

For your convenience, I computed and copied-and-pasted the annual average temperatures between 1880 and 2016 (the figures –99 mean "not yet known" but even James Hansen is almost certain than 2017 will be cooler than 2016):$\left( \begin{array}{cccccc} 1880 & -20.4 & -11.6 & -10.2 & -21. & -28.4 \\ 1885 & -31.8 & -30.6 & -33.2 & -20.2 & -11.7 \\ 1890 & -36.8 & -24.5 & -27.1 & -30.2 & -30.7 \\ 1895 & -21.6 & -14.7 & -10.9 & -27.9 & -15.9 \\ 1900 & -9.3 & -14.9 & -27.2 & -35.3 & -44.3 \\ 1905 & -27.9 & -22.7 & -40. & -43.5 & -47.4 \\ 1910 & -42.4 & -44.3 & -34.8 & -34.2 & -15.8 \\ 1915 & -10.7 & -34. & -39.5 & -26.2 & -22.2 \\ 1920 & -26.7 & -21. & -27.8 & -24.5 & -28.2 \\ 1925 & -20.6 & -9.7 & -21. & -21.5 & -36.1 \\ 1930 & -14.4 & -9.5 & -16.8 & -28.8 & -14.3 \\ 1935 & -19.6 & -15.3 & -3. & -3.5 & -3.3 \\ 1940 & 7.6 & 12.3 & 9.4 & 12.9 & 25.2 \\ 1945 & 11.5 & -3.8 & -4.9 & -9. & -8.8 \\ 1950 & -17.8 & -6.8 & 0.7 & 7.6 & -12.6 \\ 1955 & -15. & -20.3 & 3.4 & 6.8 & 3. \\ 1960 & -2.5 & 5.5 & 3. & 6.2 & -19.8 \\ 1965 & -10.2 & -4.8 & -2.1 & -7. & 6.8 \\ 1970 & 2.4 & -8.8 & 1.2 & 15. & -7.4 \\ 1975 & -1.7 & -11.4 & 18.2 & 6.9 & 16.9 \\ 1980 & 27.2 & 32.8 & 13.2 & 30.2 & 15.4 \\ 1985 & 11.8 & 19.2 & 33.5 & 40.8 & 29.1 \\ 1990 & 44.3 & 42.8 & 22.9 & 24.3 & 31.9 \\ 1995 & 45.7 & 35.2 & 48.2 & 63.7 & 41.8 \\ 2000 & 42.3 & 54.8 & 63.3 & 61.9 & 54.6 \\ 2005 & 69.2 & 63.2 & 65.8 & 53.4 & 64.4 \\ 2010 & 71.3 & 60.2 & 63.5 & 65.6 & 74.3 \\ 2015 & 86.5 & 99.2 & -99 & -99 & -99 \\ \end{array} \right)$ To prepare this $\rm\TeX$-MathJax output was simple in Mathematica. I marked the output table, right-clicked, and chose "export as TeX". The numbers in this table are exactly those captured by the graph at the top. The averages assumed that all 12 months are equally long but believe me that the errors caused by this inaccurate assumption are very small.

Just for fun. Microsoft's lead mechanical engineer has patented a foldable phone – 2x1 or 3x1 pieces. Here's an improvement: a 2x2 with four foldable pieces. A fuzzy picture worth millions of dollars and TRF readers can have it for free. The quarters of the foldable phablet look like 4-inch phones of thickness 5 mm, but this beast of total thickness 2 cm may be expanded to an 8-inch thin tablet-phone. The upper half of the vertical separating axis is cut, and those parts may be stuck together with magnets. The lower part of the vertical separator is connected by hinges near the display. The beast may also be folded around the horizontal axis near the middle, with hinges on the back side. When you pack your phone into the 4-inch thick friend, the two upper quarters of the display appear on the external sides of the thick phone and may be used as displays of the 4-inch phone on both sides.