Archive for the ‘j scott armstrong’ Category
We’ve updated the Climate Bet graph on the right with the global average temperature for August. The picture seems clear to us: there is no trend. Not surprisingly then, the projected 0.03°C per year increase that the IPCC and Mr Gore (who expected a tipping point and dangerously warming temperatures) thought was likely on the low side has been equaled or exceeded only sixteen times in the 80 month life of the bet, so far. In contrast, monthly temperatures have come in lower than Professor Armstrong’s bet on no-change for 55% of months.
Since we started monitoring the Gore-Armstrong bet back in 2008, global mean temperatures have only rarely been as warm or warmer than Mr Gore and the IPCC’s +0.03 °C per year warming forecast would have had us believe. How rarely? Well, roughly one-month-in-five, or 21% of the 77 months to date. As we’ve pointed out before, one would expect the figure to be 50% if the Gore/IPCC forecasts were unbiased. Mr Gore must be very unlucky, because the chances that so few months would turn out to be as warm or warmer than unbiased forecasts is less than one-in-eight-million.
But wait, Mr Gore and the IPCC warned us that there was more chance that temperatures would be higher than their forecasts, than that they would be lower. In other words, they claimed their forecasts were biased toward slower warming than the rapid warming they really believed would occur. We haven’t calculated the vanishingly small odds that global temperatures would turn out to be so uninclined to warm taking into account the declared downward bias in their forecasts, but we suspect that Mr Gore and the IPCC have been feeling that the Earth has let them down.
For keen followers, the latest month of The Bet, May 2014, is shown in the chart on the right.
With April’s temperature another to fall below the no-change-from-2007 line, Al Gore’s tipping point continues to elude us. The errors from the Gore stand-in forecast (the IPCC’s +0.03°C per year) are now 27% larger than the errors from Armstrong’s no-change forecasts. The Gore error’s have been so bad, that the contrary hypothesis of natural global cooling would have been a much better bet. Yes, the errors from forecasting temperatures would decline at 0.01°C per year would have been 17% smaller than Mr Gore’s errors! Click on the chart on the right for the latest data on the bet.
The average global temperature for the first quarter of 2014 was less than 0.01°C different from Professor Armstrong’s no-trend forecast. With 75 months of data now in, we wondered whether there was any evidence of bias in Professor Armstrong’s or Al Gore’s forecasts. We found that there was. Both Armstrong’s and Gore’s forecasts appear to have a bias toward forecasting temperatures that are warmer than the measured temperatures. In the case of Scott Armstrong’s no-trend forecasts, they have to date averaged 0.02C warmer than the recorded temperature. In the case of Al Gore’s IPCC alarming warming forecasts, they have averaged 0.12C warmer. We think policy makers should prefer less-biased forecasts. The updated Climate Bet graph is to the right.
With 74 months of the Armstrong-Gore bet behind us and 46 months left to go, global mean temperatures have gone nowhere. That, of course, is the no-trend forecast that is the basis of Professor Armstrong’s bet. Mr Gore on the other hand claimed temperatures would go up, dangerously. In fact, most months of the bet (57%) have seen temperatures flat or down from the previous month. Professor Armstrong’s conservative forecast has been more accurate than Mr Gore’s alarmist forecast for nearly 69% of months so far.
Mostly down. In the course of the 72 months of the bet to date, the global mean temperature fell or remained flat compared to the previous month for 41 months, or 57% of the time. (In case you’re wondering, the UAH series records only 2 occasions over this period on which the temperature did not change from the previous month.)
We’re not sure how the ups-and-downs of the global temperature over these last six years accord with what Mr Gore had in mind when he issued his warning of an immanent and catastophic “tipping point”, but we know that his chances of winning the Climate Bet against the no-trend forecast proposed by Professor Armstrong have receded as the life of The Bet has progressed.
See the chart to the right with the latest (December 2013) data.
A summary of the critique of the use of complex mathematical models for forecasting long term climate change by Kesten Green, Scott Armstrong, and Willie Soon is published in the Nongovernmental Panel on Climate Change’s Climate Change Reconsidered II: Physical Science (2013).
The reference is as follows, and links to the relevant section and to the entire NIPCC report are available from the Global Warming Audit pages of the forecastingprinciples.com site, here.
Armstrong, J. S., & Green, K. C. (2013). Global climate models and their limitations: Model simulation and forecasting – Methods and principles. pp. 14-17 in Idso, C. D., Carter, R. M., & Singer, S. F. (Eds.), Climate Change Reconsidered II: Physical Science. Chicago, IL: The Heartland Institute.
With eleven months of temperature data now in for 2013, it is highly unlikely that Mr Gore could win this, the sixth, year of The Climate Bet. For Mr Gore to win, the December global mean temperature anomoly would need to be substantially higher than any month so far in The Bet. Failing that unlikely outcome, the no-change forecast that Professor Armstrong is betting on will have been more accurate that Mr Gore’s IPCC-originated dangerous manmade global warming forecast for five of the first six years of The Bet.
With October 2013′s global mean temperature data in, we now have 70 months of evidence on the accuracy of Mr Gore and the IPCC’s alarming warming forecast of temperatures increasing at a rate of 0.03ºC per annum. If Mr Gore’s forecast were valid and unbiased, we would expect actual temperatures to be higher than his forecast roughly half of the time and lower roughly half of the time. We checked the record of The Bet. It turns out that, to date, the actual global mean temperature has been higher than Mr Gore’s forecast less than one-quarter (23%) of the time. By contrast, the measured temperature has been warmer than the no-change benchmark, Professor Armstrong’s bet, 46% percent of the time; very close to the ideal of 50%. The updated chart is to the right.
“There are no scientific forecasts of dangerous global warming” shouts a new article in the Financial Post. Readers of this blog know that already, but will likely want to read what the op-ed titled “Climate forecast: All’s well, despite what the IPCC says” by Kesten Green, Scott Armstrong, and Willie Soon has to say about the implications for government policy. It has already attracted lively discussion. The article is here.
The authors claim to provide the only scientific forecast of long-term climate, namely the naive no-change or no-trend forecast that is the basis of Professor Armstrong’s notional bet with Mr Gore. The forecast was originally published in 2009, and the International Journal of Forecasting article can be found here. The authors’ current working paper investigating possible improvements to climate forecasting for policy makers is here.