Unseasonal summer doesn't make climate models wrong

  • 04 Jul 2012, 15:44
  • Verity Payne

Everyone knows we Brits love to grumble about the weather. Now climate skeptic commentators are exploiting our love of a good weather-related moan to suggest projections about long-term climate can't be right. But seasonal projections - a developing area of climate forecasting - are quite different beasts from more mature techniques like short term weather forecasting and long-term climate modelling. We've taken a look.

Weather grumbling

The Met Office comes in for a lot of criticism when Britain's suffering bad weather. In part, this is probably more of that love of grumbling we were talking about - as Rob Varley, Met Office operations and services director told the  Telegraph earlier this month:

"There is a link between weather that people don't want and criticism of what we do."

Over recent weeks, skeptic commentators  Richard  Littlejohn Christopher Booker and  Matt Ridley, along with lobby group the  GWPF have all focused on the terrible weather in order to return to one of their favourite arguments.

The argument goes: the Met Office's seasonal forecast - for April, May and June - didn't predict the record rainfall we've experienced, so this calls into question projections of climate decades or centuries into the future.

Says Littlejohn:

"[W]hy should we believe long-term 'climate' predictions from global warming scaremongers like the Met Office when they can't even get the weather forecast right from one week to the next?"

Climate is what you expect. Weather is what you get.

First up, this argument confuses climate and weather. Weather is the short-term - day to day, week to week - change in atmospheric conditions. Climate, on the other hand, is the long-term average of weather conditions. To work out how climate is changing scientists tend to use records spanning at least  30 years.

The website Climate Bites has some useful analogies for the difference between weather and climate, including:

"Drawing conclusions about climate by looking at the weather is like saying 'I lost 2 lbs yesterday!' Every veteran weight-watcher knows that one day means nothing. It's the long term trend that counts."

And this video is useful:


Having said this, there is some interesting new research in this area which suggests a different story. Physicist Shaun Lovejoy and colleagues suggest that what we traditionally think of as climate - the predictable average of weather - is actually 'macroweather'.

Lovejoy proposes climate fluctuations act on scales greater than 10 - 30 years, and may be controlled by less predictable drivers, such as deep ocean currents. This is quite new stuff, but it's worth keeping an eye on - if true, it will have implications for uncertainty in climate projections. More from Lovejoy lower down.

Prediction is difficult. Especially about the future.

This fundamental distinction between weather and climate means that forecasting the two are rather different propositions, although the models used to simulate both weather and climate are basically the same.

The main thing that determines a day to day weather forecast is the data that gets put into the model - observations of recent weather such as temperature, wind speed, wind direction, and how much rain or snow there's been, from as many locations as possible. These data are known as the 'initial values'. Computer models predict how the weather might evolve over the following days, so to a great degree the quality of these measurements decides the quality of a weather forecast.

Any small errors in these observations can quickly escalate as time moves on, so the further into the future a model forecasts, the less accurate it becomes. (You can see for yourself how accurate the Met Office's recent short term weather forecasts have been here.)

Forecasting the exact weather many years or decades ahead is simply not possible, so these initial values are not so important for climate projections. Instead, in climate modelling, the aim is to work out long-term climate trends caused by likely future changes to climate drivers - things like how much sunlight there'll be, how heat moves through oceans, or greenhouse gas levels.

How well the model can simulate changes to climate due to these climate drivers determines the quality of climate projections. By repeating model simulations many times, scientists can get an idea of the range of likely outcomes caused by future changes in climate drivers. But this is quite a different thing than trying to do a weather forecast for the next hundred years.

Seasonal projections are a work in progress

Seasonal forecasts sit somewhere between climate and weather forecasts, and only give the likelihood of experiencing different types of conditions compared to normal - for example how 'dry', 'near-average' and 'wet' it might be. For April, May and June the Met Office said:

"The [seasonal] forecast for average UK rainfall slightly favours drier-than-average conditions for April-May-June as a whole, and also slightly favours April being the driest of the 3 months [...] The probability that UK precipitation for April-May-June will fall into the driest of our five categories is 20-25% whilst the probability that it will fall into the wettest of our five categories is 10-15%"

The fact that the 'slightly favoured drier than average' conditions didn't materialise doesn't really make the seasonal forecast 'wrong'.The forecast gave a 10-15 per cent chance of there being very wet conditions; it was indeed very wet.

Essentially, if you want to call a seasonal forecast 'right' or 'wrong', you're usually going to be able to do so. The Met Office describes this as "the absurd situation that a single probabilistic forecast is always 'right'. Simply, there is no way of verifying a single probabilistic forecast." At the same time, detractors of the weather forecasters will usually be able to find part of the forecast that doesn't pan out.

Met Office Head of Climate Impacts Richard Betts told his twitter followers:

"[S]easonal forecasting is very hard, but we're giving it a go in order to learn and improve"

So the claim that because seasonal forecasts are 'wrong' we shouldn't trust long-term climate projections is also based on a misunderstanding of how to read the Met Office seasonal forecasts.

All models are wrong, but some are useful

The uncertainties associated with long-term climate projections - to 2100, say - are likely to get larger. For example, Lovejoy told us that if his new definition of climate is right;

"It means that [climate model] projections past 10-30 years will be less reliable - the missing mechanism could make anthropogenic effects either "better" or "worse" depending on feedbacks. I don't think any of this brings into question the reality of anthropogenic warming, however it does increase the range of uncertainty associated with future scenarios (chiefly beyond 10- 30 years)."

But does a large uncertainty range in climate model projections mean we should just ignore the projections? Whether we chose to accept or ignore model projections is a risk, as Professor Peter Muller, University of Hawaii, explains:

"Not doing anything about the projected climate change runs the risk that we will experience a catastrophic climate change. Spending great efforts in avoiding global warming runs the risk that we will divert precious resources to avoid a climate change that perhaps would have never happened. People differ in their assessment of these risks, depending on [things like] their values. To a large extent the discussion about global warming is about these different risk assessments rather than about the fairly broad consensus of the scientific community."

So we can choose to accept or reject action based on climate projections. But making the choice should probably be informed by more than what the weather's been recently.

Email Share to Facebook Stumble It
blog comments powered by Disqus