Archive for the ‘forecasting’ Category
Last month we noted that if temperatures remained at around the same level as they averaged for the first three months of 2016, Mr Gore could win the Climate Bet. The figure—global average temperature anomaly—was +0.7°C.
We didn’t fancy his chances given that the figure is a high for the satellite record and is associated with a strong El Niño weather pattern. The warmth bringing El Niño weather pattern is followed by cooling La Niña weather. One month on, however, the April 2016 figure turned out to be +0.71°C. We imagine Mr Gore must be overjoyed!
With 20 months of the 10-year bet to go, we still don’t fancy Mr Gore’s chances that 0.7°C temperature anomalies will continue. We’ll keep you posted.
While, after 99 months of the Climate Bet, Mr Gore’s forecast errors are 37% larger than Professor Armstrong’s, it is mathematically possible for Mr Gore to win. For that to happen, however, the global average temperature anomaly would have to stay around the average of the first three months of this year, +0.7°C. We will keep you posted!
We’ve had to adjust the Climate Bet chart to make room for the February 2016 UAH global average temperature anomaly of +0.83°C above the 1981-2010 average. For five months in a row now, Mr Gore and IPCC’s warming projection was more accurate than Professor Armstrong’s no-change-from-2007 forecast. The last time Mr Gore got such a run was in 2010, when the IPCC warming projection was more accurate for the first 10 months of the year.
Overall, however, the errors of IPPC projection are still as much as 40% larger than the errors from the evidence-based forecast of no change.
Mr Gore’s chances of winning the bet must, nevertheless, have improved with the latest figure. To put the data into perspective, if the temperature anomaly remained at or above .437°C for the last 22 months of The Bet—to the end of 2017—Mr Gore would win.
Perhaps Mr Gore will he change his mind and decide that he would like to put some of his own money at stake. He should be aware that the .437°C figure has only been equaled or beaten 7 times over the 98 months of The Global Warming Challenge to date, but that statistic will presumably carry little weight for those who, like Mr Gore, believe in the coming of a “tipping point”.
See the updated Climate Bet chart to the right.
The prestigious Nature magazine on 24 February 2016 published online an article recognising the fact that (looking back) global average temperatures have been trend-less for at least the last 15 years. That’s right, Mr Gore, despite greatly increasing concentrations of atmospheric carbon dioxide, the various measures agree that there is still no tipping point!
David Whitehouse, in a 26 February article in the U.K.’s The Spectator, suggested that Nature‘s article “ought to have been front page news – and might have been, had it suggested that global warming was worse than we had thought.” Whitehouse’s article, with link to the Nature paper, is available here.
The short answer is, most likely it is too late for Mr Gore to win the bet with Professor Armstrong. Here’s why…
Of the eight years of the bet so far, Professor Armstrong’s no-change forecast has been been more accurate than the Gore/IPCC “dangerous” warming forecast in seven. Looking at individual months, the no-change forecast has been more accurate for 72 of 96, or 75%. Overall, the errors of Mr Gore’s forecasts to date are nearly 42% larger.
To tip the balance of forecast errors back in Mr Gore’s favor, the temperature anomaly would need to average close to the Gore/IPCC +0.03ºC p.a. trend line shown in the updated chart to the right.
For those of you who arrived late or who would like a refresher, brief background to the Armstrong-Gore bet is provided at the link in the right column labelled “Challenge,” or here. (We recently extended the description of the background to the Challenge.)
The 93 months of the 120 month (10-year) Climate Bet so far has witnessed 45 months in which the global average temperature anomaly increased from the previous month, and 46 months in which the global temperature fell. This pattern, or lack of it, is of course consistent with the Green, Armstrong, and Soon (2009) evidence-based no-change forecast that is the basis of Professor Armstrong’s notional bet with Al Gore. For the latest data, click on the chart to the right.
An up-tick in temperature anomalies in June saw Mr Gore and the warming scenario score the first win against the no-change forecast since January of 2013, nearly two-and-a-half years ago. The outlook for the dangerous warming scenario remains bleak, however. Over the 7.5 years of the Armstrong-Gore Bet so far—we have now past the ¾ mark—the errors that have arisen from projecting temperature to increase at a rate of 3°C per century are more than 50% larger than the errors from the no-change forecast.
Is it really possible that the simple no-change forecast of 21st Century temperatures is better than the IPCC projections from expensive and complex computer models? Yes, it is. That conclusion is consistent with the evidence presented by Kesten Green and Scott Armstrong in their recently published review of evidence on the effect of complexity on forecasting. They found that using complex methods increases forecast errors relative to the forecasts from simple methods that decision makers could understand by 27% on average. We expect that the results of The Climate Bet will increase that average.
For the latest data from UAH and the progress of the bet, see the new chart to the right.
The Heartland Institute’s Tenth International Conference on Climate Change (ICCC 10) took place in Washington D.C. on the 11th and 12th of June. Scott Armstrong presented a talk based on research with Kesten Green. Slides of their talk can be downloaded by clicking here. A flyer, summarising their evidence on climate forecasting, with links to relevant papers is available, here.
For this who missed the conference or would like to catch talks that they missed, videos of the ICCC 10 talks are now available online here. Scott gave his talk in a session with Anthony Watts and Roy Spencer. Video of their excellent session is here.
While Mr Gore with his expectation of a “tipping point” and the IPCC with their dangerous warming projection will no doubt be surprised at how low global temperatures have been running, Professor Armstrong with his scientific forecast of no long-term trend in temperatures will not.
The Armstrong-Gore bet has now been running for 7 years and 5 months (89 months) now, and the average global temperature anomaly as calculated from satellite measurement by the UAH team has been 0.12°C. That figure compares with the 0.17°C average for the base year of the bet, 2007. That’s right, the average global temperature over the nearly 90 months since the beginning of the bet has been lower than the average for year the bet is based on.
The non-tipping point that we have been experiencing for more than seven years leaves Mr Gore’s bet out in the cold. His average absolute error to date is 0.22°C. That figure is 55% greater than the error of Professor Armstrong’s scientific forecasts. Yes, the scientific method does work, and can be relied upon ahead of the opinions of experts (even those of scientists) every time!
For the latest temperature data, click on the chart to the right of the screen.
The benchmark global temperature data from the researchers at UAH-Huntsville have been adjusted to compensate for drifting in the positions of the satellites that take the readings and other improvements in the measurements and calculations. The improvements in the data series must be disconcerting for warming alarmists such as Mr Gore and the IPCC: dangerous warming and a “turning point” are nowhere to be seen. We hope they are relieved that there is (even less) reason to believe the Earth is in danger and that governments will realise the folly of policies to reduce carbon dioxide levels.
Our chart of The Bet to April 2015 is shown to the right of the page, as usual, using UAH’s revised series. The picture is clear, but for this who prefer numbers, here is a very small one: 0.000000000005. That number (which is roughly equal to 1-divided-by 214 billion) is the probability that temperatures would have equaled-or-exceeded Mr Gore and the IPCC’s 0.03°C per annum warming projection as few or fewer times as the 13-out-of-88 months of The Bet so far that they have done so… if their projection were unbiased.
For more information on the UAH data revisions, see the description by Spencer, Christy and Braswell, here.