Climate change

March 11, 2011

A toy model for forecasting global temperatures – 2011 redux, part 1

Filed under: Climate Change, Global Warming — buildeco @ 1:14 pm

by Barry Brook

A little over two years ago, I wrote the following post : How hot should it have really been over the last 5 years? In it, I did some simple statistical tinkering to examine the (correlative) relationship between global temperatures and a few key factors, namely greenhouse gases, solar irradiance, and ENSO. In the next couple of posts, I’ll update the model, add a few different predictors, and correct for temporal autocorrelation. I’ll also make a prediction on how global temperatures might pan out over the coming few years.

In the 2008 post, I concluded with the following:

To cap of this little venture into what-if land, I’ll have a bit of fun and predict what we might expect for 2009. My guess is that the SOI will be neutral (neither El Niño or La Niña), the solar cycle 24 will be at about 20% of its expected 2013 peak), and there will be no large volcanic eruptions. On this basis, 2009 should be about +0.75, or between the 3rd and 5th hottest on record. Should we get a moderate El Niño (not probable, based on current SOI) it could be as high as +0.85C and could then become the hottest on record. I think that’s less likely.

By 2013, however, we’ll be at the top of the solar cycle again, and have added about another +0.1C worth of greenhouse gas temperature forcing and +0.24 of solar forcing compared to 2008. So even if 2013 is a La Niña year, it might still be +0.85C, making it hotter than any year we’ve yet experienced. If it’s a strong El Niño in 2013, it could be +1.2C, putting it way out ahead of 1998 on any metric. Such is the difference between the short-term effect of non-trending forcings (SOI and TSI) and that inexorable warming push the climate system is getting from ongoing accumulation of heat-trapping greenhouse gases.

So, now that we have data for 2009 and 2010, how did I do? Not too bad actually. Let’s see:

1. After bottoming out for a long period, the 11-year sunspot cycle has restarted. So much for those predicting a new Maunder Minimum. By the end of 2010, we had indeed reached about 20% of the new forecast maximum for cycle-24 (which is anticipated to be about half the peak value of cycle-23).

2. We had a mild El Niño in 2009 and early 2010, before dipping back into a strong La Niña. See here.

3. There were no large equatorial volcanic eruptions. The best we got was Iceland’s Eyjafjallajökull (don’t ask me to pronounce it), which actually helped climate change a little bit by stopping flights over Europe for a week.

4. 2009 was ranked as the 5th warming on record. I had ‘forecast’, based on my toy model, that it would be somewhere between 3rd to 5th. I said that 2009 would be about +0.25C hotter than 2008; the real difference was ~ +0.15C (based on the WTI average index data). This was followed up by 2010 equalling with 2005 as the hottest year on record. Pretty much right in line with my guesstimate.

5. Anthropogenic GHGs continue to accumulate; atmospheric CO2 concentrations built up by 1.9 ppm in 2009 and 2.4 ppm in 2010. That forcing ain’t going away!

I still stand by my 2008 prediction of us witnessing record-smashing year in 2013… but I’ll have to wait another couple of years to confirm my prognostication. However, I’m not going to leave it at this. There are a couple of simple ways I can improve my toy model, I think — without a lot of extra effort. Doing so will also give me a chance to show off a few resampling methods that can be used in time-series analysis, and to probe some questions that I skipped over in the 2008 post.

In short, I think I can do better.

In Part 2, I’ll describe the new and old data streams, do some basic cross-correlation analysis and plotting, bootstrap the data to deal with autocorrelation, and look briefly at a method for assessing a model’s structural goodness-of-fit.

In Part 3 I’ll do some multi-term fitting to data going up to 2005, and use this model to project the 2006 — 2010 period as a way of validating against independent data (i.e., that not used in the statistical fitting), then re-fit the best predictive model(s) to all the data, and make a forecast for 2011 to 2013 inclusive. The big catch here will be getting the non-CO2 forcings right.

Stay tuned


1 Comment »

  1. Should we act on the climate change information we have been fed?

    I think this little post of mine illustrates that dilemma very well.



    Comment by rogerthesurf — March 11, 2011 @ 4:24 pm

RSS feed for comments on this post. TrackBack URI

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Create a free website or blog at

%d bloggers like this: