Saturday, February 04, 2017

Vostok: warming or cooling by 2 °C per century has been frequent enough and natural

An hour ago, I was intrigued by the apparent asymmetry of warming and cooling episodes in various graphs. In particular, if you look at the temperature graphs of the Vostok ice core, you will immediately see that the warming episodes were often much faster and speedier than the cooling episodes. Most of the time, it was cooling. A minority of the time has experienced warming but it was appropriately more intense. It seems that it's easier to destroy ice sheets than to build them etc.

After I have spent some time by quantifying these asymmetries in the Vostok data – and recent GISS temperatures – with various choices of the time delay etc., I decided to look at other properties of the Vostok raw data. Note that the ice is some 3.31 kilometers thick. The deeper you get, the more distant past is revealed by various gases dissolved in the ice. The deepest ice over there is 422,766 years old. Near the surface, one meter corresponds to some 17 years; near the bottom, one meter corresponds to some 600 years. The temperature is determined from the concentration of deuterium in the ice.

While the graph of the temperatures is mostly continuous, I was rather shocked by some medium-term temperature changes. To look at them, I defined an array with 422,767 entries and calculated the temperature X years ago (well, before the year 2,000 when the data was published) for X between 0 and 422,766 by linear interpolation (which understates the variability if it does something). Then I looked at the temperature changes over all 100-year (and also 50-year, 150-year etc.) intervals and drew their histogram.

The histogram for 100-year temperature jumps in the Vostok ice core looks roughly like the function $$2^{-|\Delta T / 0.2|}$$. So every time you increase the jump $$\Delta T$$ by 0.2 °C, the number of moments when the jump is this high gets halved. Consequently, there are lots of moments (perhaps $$1/32$$ if you take my rough description seriously) when the temperature change over 100 years exceeded 1 °C. And it was a completely natural temperature change.

Try to look at the raw data. In the first column, you see the depth in meters, OK? It's between 0 and 3310 meters. Let me label the rows by this depth. Now, have a look e.g. at the rows 198-202. Between the years 8226 and 8043 BP (i.e. 6226 and 6043 B.C.), the temperature anomaly went from –0.87 °C to +2.06 °C and back to –0.64 °C. In less than two centuries, the temperature increased by almost 3 °C and then dropped by more than 2.5 °C again.

If you believe this data at this level of detail, the rate of warming (or cooling) was three times as fast as what we saw in the 20th century. And there were no SUVs then.

Or look at lines 85-87. The temperature anomaly dropped from +0.58 °C to –1.97 °C, i.e. by more than 2.5 °C, in less than 100 years, between 847 and 760 B.C.

Rows 74-75: over 2 °C of warming in just 40 years around 300 B.C. Lines 19-20: 0.98 °C of warming from 1580 to 1603 AD, just in 23 years! There are lots of examples like that. Let me add two more ancient ones: Lines 1174-1175 (or 1176): over 1.2 °C of cooling in 67 years (or 2.1 °C of cooling in 128 years), some 82500 B.C. Lines 1847-1848 (or 1846-1849): 2.87 °C of warming in 54 years (or 3.09 °C of warming in 162 years) around 127400 B.C.

The main question is whether this detailed data may be trusted at the centennial scale. If it can, the centennial averaged temperatures at a place, in this case a place in Antarctica, may very well change by more than 2 °C in one century. The causes of these warming and cooling episodes were almost certainly natural. It's almost completely normal for the temperature to change by more than 2 °C per century. What we were getting in the 20th century is much more modest.

Even if you could set all the human influences affecting the climate to zero, the natural variability would continue. And it would sometimes produce more than 1 °C. For a random 100-year reference period, the probability is roughly several percent that the temperature change over the period exceeds 1 °C (with either sign), according to the Vostok ice core.

This natural variability is a background that should have similar consequences for our thinking about the risk as the natural radioactivity. Even if the fossil fuels etc. were responsible for 1 °C or 2 °C of "extra warming" in the 21st century, the temperature change in the century would probably be something that the Earth experiences several times in 10,000 years, anyway. The mankind – and most other species – have surely survived such episodes in the past. It's much easier to deal with them today.

But this isn't the most important point, I think. What I find more important is that the Vostok data indicate that there simply exist natural reasons that may change the temperature by more than 1 °C and perhaps more than 2 °C per century rather often and these effects, whatever they are, therefore may be responsible for the 20th century "global warming", too, or for most of it. Also, 1-degree or 2-degree warming episodes were often followed by similarly abrupt cooling (or vice versa). We can't really exclude this possibility.

The 20th and 21st century may really be analogous to the years between 6226 and 6043 B.C. Back then, the temperature went up by 2.5 °C in the first century and returned back down by 2.5 °C in the following century. How confident you may be that the 21st century won't experience similar cooling as the cooling between 6091 and 6043 B.C. which followed a century of rapid warming? I think that you simply can't be more than 99 percent confident because in 1% of cases or so, the Earth managed to produce a significant cooling like that.