Trend and Fluctuation (One More Time)

A recent comment says:

The Gistemp temperature anomaly for June 2015 was 0.79 degrees Celsius.
The Gistemp temperature anomaly for February 2016 was 1.32 degrees Celsius.
In 8 months the temperature anomaly increased by 0.53 degrees Celsius.

The question is, how much of the 0.53 degrees Celsius increase was caused by man-made global warming?

Let me propose an analogous question:

The temperature on March 1st reached 2°C. The temperature on March 2nd reached 18°C. In 1 day the temperature increased by 16°C. How much of the 16°C increase was due to the progress of the seasons from winter toward spring and on to summer?

Maybe it’s obvious, even to the author of the aforementioned comment. But just in case, the answer is: the day-to-day fluctuation in temperature swamps the amount of trend change in a single day.

Only a fool would claim that most, or even a sizeable fraction of, the 16°C rise from one day to the next was due to the passing of the seasons. Only a bigger fool would ask the question with a straight face. Only the biggest fool would attempt to use the day-to-day fluctuations in temperature as evidence that there’s no seasonal trend and no reliable evidence that summer will be hotter than winter.

Global average temperature data from NASA GISS looks like this:


The black line follows the monthly average temperature anomaly. The thicker red line shows the trend estimated by linear regression. The blue circles mark the values in June 2015 and February 2016.

Fluctuations come and go, but over the long haul they average out to nothing. Trends persist, so over the long haul they can accumulate to large changes in spite of the fact that in the short term they amount to very little. That’s why, just since 1970, the net trend change in the NASA GISS data is 0.84°C, considerably bigger than the short-term change from June 2015 to February 2016 which is due to fluctuation.

Global warming is about the trend, not about the fluctuations. The progress of the seasons is about the trend, not the fluctuations.

If the trend continues at the present rate, then by January 2100 the net warming due to trend will be 2.34°C above the 1970 value. And of course, it’s possible the average rate over the remainder of this century will be higher than it has been recently, especially if warming triggers feedback in the carbon cycle (releasing natural stores of CO2 or CH4).

Let’s avoid that.

Let’s also stop treating claims or questions which conflate trend and fluctuation in the most ludicrous way, as being serious claims or questions, let alone worthy of “debate.”

This blog is made possible by readers like you; join others by donating at Peaseblossom’s Closet.

5 responses to “Trend and Fluctuation (One More Time)

  1. scottdenning156684402

    As always, Tamino’s discussion of signal:noise and trend vs fluctuations is clear and illuminating. But the extrapolation in time (to 2 C of warming by 2100) ignores the cause-and-effect relationship to changes in atmospheric CO2, which is not linear in time but is rather related to the time integrated combustion of coal, oil, and gas.

    When we account for projections of economic growth, population growth, and changes in the energy supply, as IPCC does, we obtain something like twice the warming by 2100 as the extrapolation of the temperature trend over the past 50 years. In other words, we have every expectation that the trend will substantially accelerate based on very simple cause and effect relationships among emissions, concentrations, radiation, and temperature. The future will not be like the past, unless very strong policies are enacted to slow the growth of fossil fuel combustion, which is of course precisely the goal of the Paris agreement last year.

  2. The Met Office forecast for 2016-2020:

    This forecast also suggests global temperatures over the next five years are likely to be well within, or even in the upper half, of the range of warming expected by the CMIP5 models, as used by the Intergovernmental Panel on Climate Change.

    Is there any chance the trend for the first 2 decades of the 21st century will be ~.2 ℃ per decade?

  3. Technically, one could reply to the original question with a result even if it’s not meaningful. I’m gonna do the math here, even if it’s garbage in, garbage out, just to consider the question technically answered.

    The trend of 0.84°C over 46 years would average 0.0183°C per year or 0.0122°C over that eight month period, which is 2.3% of the 0.53°C difference in anomaly values.

    Feel free to check my math or dismiss it out of hand if it’s based on wrong assumptions.

  4. Maybe it’s useful to call out the rhetorical trick with this kind of question. Something like this:

    The question is interesting because it is in the form of a ‘gotcha’, which in turn sets up a ‘lie of context’. I’ll explain those.

    Most people ask questions to learn. However sometime people ask questions in order to score debating points – those are ‘gotcha’ questions.

    In this case the question is chosen so that the response highlights the fact that internal variability (weather) is much larger than climate change on short timescales. This is true, and well know, but the aim is to get the reader to infer that since the weather signal is larger than the climate signal over short periods, climate change is not a problem.

    This is a lie of implication or a lie of context. The reader is being given correct but incomplete information, on which they are then drawing a false conclusion. To draw a correct conclusion, the reader needs to also know the difference between weather and climate, and that the weather signal is largely random, while the climate signal is monotonically increasing and overwhelms the weather signal over longer periods. The questioner relies on the reader not having this information and therefore drawing a false conclusion.

  5. Many who dispute the reality of human-induced climate change have taken lately to claiming short-term fluctuations are all that exists. (Some admit there may also be long-term fluctuations, but they are cyclical and will reverse, and in any case we can’t do anything about them.)

    Thus, a series of recent papers has explained the increase in seasonal Antarctic sea ice in terms of fluctuations in ocean currents and winds around Antarctic (it’s quite a bit little more complex, but I hope that isn’t a misleading capsule summary). Climate science deniers had been using Antarctic seasonal sea ice as an argument supposedly “disproving” claims of a warming world. The recent papers show that no, it’s simply a case of temporary fluctuations that are for a time overwhelming a slow but steady long-term trend–a trend which will win out in the end, just as the progress from winter to summer will certainly trump the daily up-and-down fluctuations of temperature or precipitations.

    Yet deniers will seize even this truth, claiming that since short-term variation can, for a time, overcome longer trends, that must mean the longer trends, if they even exist, are inconsequential. Day-to-day and year-to-year noise is always more noticeable than the steady march of the underlying signal. The import of the signal is a difficult message to convey to those who are proudly innumerate.