Natural Variability, Attribution and Climate Models #2

Introduction to Using Simple Physics or Complex Models on the Challenging Question of Attribution

In #1 we looked at some examples of natural variability - the climate changes from decade to decade, century to century and out to much longer timescales.

How sure are we that any recent changes are from burning fossil fuels, or other human activity?

In some scientific fields we can run controlled experiments but we just have the one planet. So instead we need to use our knowledge of physics.

In an attempt to avoid a lengthy article I’m going to massively over-simplify.

“Simple Physics”

Some concepts in climate can be modeled by what I’ll call “simple physics”. It often doesn’t look simple.

Let’s take adding CO2 to the atmosphere. We can do this in a mathematical model. If we “keep everything else the same” in a given location we can calculate the change in energy the planet emits to space for more CO2. Less energy is emitted to space with more CO2 in the atmosphere.

The value varies in different locations, but we just calculate it in lots of places and take the average.

As less energy is leaving the planet (but the same amount is still being absorbed by the sun) the planet warms up.

In our model, we can keep increasing the temperature of the planet in our model until the energy emitted to space is back to what it was before. The planetary energy budget is back in balance.

So we’ve calculated a new surface temperature for, say, a doubling of CO2.

Hopefully, the idea is fairly simple.

This simple model has ignored lots of complexities - lots of ways in which the climate responds - but it gives some basic insights.

Basic insights are valuable.

The other important point about this simple model is we can test how sensitive the response is to minor changes. If a small change in a starting condition gives a massively different response then we have a problem. For increasing CO2 under this “simple example” we don’t have this problem. The results aren’t particularly sensitive to minor details.

In this simple example, we find something fascinating.

While the surface of the planet warms, the stratosphere (the upper atmosphere) cools. That’s what the mathematical model shows.

It also turns out that observations of the stratosphere show this same cooling.

So we have a simple model reproducing two key observations. If we try the same test with increasing solar radiation we don’t get a cooling stratosphere. It’s definitely evidence pointing towards the obvious (plus, see notes).

“Complex Physics”

In the simple model we did one thing - increase CO2 in the atmosphere - and the temperature increased. Clearly other stuff will change. It gets hotter so there’s more water vapor (warm air holds more moisture) and so more rain. Ice melts. Perhaps the atmospheric circulation changes? Perhaps the ocean circulation changes?

Enter the GCM, alternatively standing for global climate model or general circulation model.

Here, the world is divided up into a large grid and a supercomputer solves a large number of equations in each grid point - temperature, wind in each direction, water vapor, rain, etc, etc, etc.

Then the “time” is moved forward a few hours and the equations are recalculated.

The grid is typically something like 100km x 100km for the latest GCMs.

Climate models do some things well and some things badly. They can reproduce some aspects of climate.

One way that climate scientists try to figure out “attribution” is to run a model, or a bunch of models, over the recent past with and without the CO2 changes and look at the difference. If the one with our CO2 increases is a good match, and the one without is a bad match, that is considered evidence for attribution.

How convincing this is depends on how confident you are that climate models are good at their job.

We’ll look at this more in the next article.

---

Notes

We’ve had satellites in space for over 40 years measuring incoming solar radiation. There’s a cycle but no overall trend: