If you took an average output of multiple climate models, it would predict that the start of this century would have seen a strong warming trend. Instead, the planet warmed relatively slowly over this time.
When models and reality disagree, it can tell us about two things: the models and reality. So far, analysis has seemed to come down on the side of reality. Evidence has indicated that one of the contributors to this century’s climate has been small volcanic eruptions; another suggests that a run of La Niña years has helped hold temperatures down.
Now, a new study is out that turns the focus on the models. It finds no evidence that the models are biased toward predicting higher temperatures and instead suggests that their biggest issue might be in how they handle large volcanic eruptions.
In an intriguing study, the researchers compared the trendlines for a variety of models with real numbers over the last century. They too the models, started in 1900 and rant the clock forward. Then added in the real data to see how well it fit.
What they found was that the real numbers all fell within the error bars of the models. So there was no systemic bias in the models to over estimate temperature change.
Essentially, each model has intrinsic errors around the trendline and the trends always stay within the error bars.
But they did find several periods where the trendline came closer to either the high or low error range for short periods. The temperature would not rise for 10-15 years, getting closer to the low range and then shoot up faster for 10-15 years.
So, while the average trend (and its error bars) has been going up over the last 100 years, there were real periods of slower and faster warming.
These times of slower warming appear to correspond to large volcanic activity. The volcanoes send up large amounts of dust and gases which will affect climates. But as they cycle out of the atmosphere, warming returns.
What seems to be happening is that the models may not correctly model the effects of volcanoes. They overestimate the cooling effects of volcanos, modeling a lower increase in temperature. But then they correspondingly overestimate the heating effects as the volcanic impacts lessen.
They get the average correct but overcompensate because they give too much impact to volcanoes. Better modeling of vocanoes would not change the overall average trends. It would just reduce the error bars.
What this means is that when you look at any 10-15 year trend, the explanation will be mostly due to temporary chaotic’ and unpredictiable conditions, like volcanoes. But these disappear when longer time periods are examined because the system reverts to normal processes.
And by modeling these chaotic events better – ie volcanoes – they will not chaneg the overall trend, only reduce the error bars.