In the last post we talked about trend and noise, how trend reflects climate change while noise doesn’t, and the need to separate the two in order to figure out how global temperature has warmed, not just fluctuated. Now, let’s talk about the patterns of global temperature change since 1880, and what might have been the causes. I say “causes” rather than “cause” because there are many things that can affect global temperature.
One of the ways to separate trend from noise is to smooth the data. I’ll use two different smoothing methods. One is a lowess smooth, and it gives this — the smooth is shown as a thick red line, the data (annual average temperature anomaly from NASA) as black dots connected by a line:
The other method is based on the fact that, from on the lowess smooth, we can see that the overall pattern can be divided into four distinct time spans. From 1880 through 1911 it declined slightly, from 1912 through 1942 it increased, from 1943 through 1971 it leveled off (maybe even declined a bit), and since then it’s been rising rapidly. During each episode it’s close to following a straight line. When you analyze the data deeply, you can show that from a statistical point of view that’s about all that we can prove. It might be showing more complex behavior than just following four straight lines, in fact it almost certainly is. But that’s all we can demonstrate with confidence, statistically speaking.
So, our other smoothing method will be to fit four straight lines to the data, which must connect together at their endpoints. We’ll call it a “continuous piecewise linear” fit because it consists of pieces, each of which is a straight line, and they have to meet at their endpoints. The times at which the lines change are only estimates (that’s the way data analysis is!). It gives us this:
The smooth fits give us an estimate of the climate signal, with the effect of noise greatly reduced. We can therefore use it to estimate how fast global temperature is changing in terms of climate change rather than just fluctuations.
If we use the lowess smooth to estimate the warming/cooling rate we get this:
The solid red line is the estimated warming rate, the dashed red lines show an estimate of how uncertain our estimate is. Keep in mind that both the rate, and its uncertainty, are only estimates.
The piecewise-linear fit consists of straight lines, and for each one the rate is constant. So it’s no surprise that the estimated rate consists of four constant values:
Something else to bear in mind is that for the piecewise-linear fit, the estimated rate is actually the average rate throughout the time span it covers, not necessarily the instantaneous rate at a single moment of time.
We can compare our two estimates of the warming rate, to see how well they agree:
The agreement is quite good. There are differences of course; the piecewise-linear fit must, by its very nature, be constant during each “piece” while the lowess-smooth fit, by its very nature, can’t change suddenly from one moment to the next.
What we can be confident of is that there are at least four episodes of different warming rates.
What caused those differences?
During the first interval, before 1912, there was very little influence from man-made factors (“anthropogenic” climate forcing). There was some; CO2 had already increased above its value prior to the industrial revolution, and methane (CH4, another greenhouse gas) was also higher. But by and large the changes were due to natural factors.
The dominant natural factor in that period is volcanic eruptions. A very large volcanic explosion puts a lot of junk in the air, and if the explosion is big enough it can reach the stratosphere and hang around in the atmosphere for several years. Part of what it spews forth is sulphur compounds, which can form sulfate aerosols. These tend to be bright, and to scatter some of the incoming sunlight right back to space so it never reaches Earth’s surface and doesn’t help warm the planet. When incoming solar energy is reduced, that tends to cool off our climate.
There were several large eruptions in that time span, most notably the Krakatoa volcano in 1883, an explosion so massive it was heard from 1,000 miles away! But it wasn’t the only one (just the largest), and the collective effect of those volcanic explosions helped cool of the earth slightly in the first period we’re discussing.
After 1912, few volcanos helped cool the earth. But greenhouse gases, especially CO2, were building up even back then. There was also a slight increase in the output of the sun itself. The warming influence of greenhouse gases and increased solar output, together with a “recovery” from the cooling influence of prior volcanic eruptions, brought about global warming.
From around 1943 up to 1970, temperature leveled off. This is most likely due to sulfate aerosols, but not from volcanic eruptions. This was a period of massive industrial expansion after World War 2, and that required burning a lot of fossil fuel, which at the time meant a tremendous amount of coal. Back then we were burning “dirty coal,” which emits a lot of sulphur. So, even though greenhouse gases were building up, the cooling influence of man-made sulfate aerosols drove temperatures down, and the two together meant little change in global temperature.
But there’s another effect from man-made sulphur in the air. It also leads to the formation of sulphuric acid in the air, which brought about the problem of acid rain (and other problems too). By 1970 this was becoming a serious enough environmental problem that industrialized nations in North America and Europe passed laws requiring the removal of sulphur from coal, or the removing of sulphur emissions from burning it. This meant that the cooling influence of man-made sulfates was greatly reduced.
With the cooling checked, the greenhouse gases took over. They’ve been the dominant factor in global temperature change since the 1970s, and we can see that in the temperature data.
Climate scientists use computer model simulations to compare how natural forces alone have tended to change climate, to how all forces — both natural and anthropogenic — have. It the latest report of the IPCC (Intergovernmental Panel on Climate Change) they graph the comparison directly:
The bluish band shows the range suggested by natural factors only, the pinkish band shows what we get with all factors including human influences. The thick black line outlines what we’ve actually observed. It’s quite clear that without human influences there’s no explanation for the changes we’ve see, but when we include the impact of mankind the computer model simulations match the observed temperature changes well.
Of course there’s a lot more to the story, and a lot we don’t fully understand about what’s influencing climate. But we also understand a lot, and there are certain things that are beyond doubt. Gases like CO2 and methane are potent greenhouse gases. They’re on the rise, and have been for a long time. There are also “feedbacks” in the climate system. For example: when the atmosphere warms it can hold more water vapor (that’s basic physics), and there’s more in the air now than there used to be, and water vapor is also a greenhouse gas — so, more CO2 means higher temperature means more H2O means: even higher temperature.
We’re currently on a path to global temperatures this century that haven’t been seen on Earth for a very, very, very long time. The consequences will be profound. But, that is a topic for another day.