Dr James Carruthers is a climate scientist at Newcastle University. As the Premier League season reaches its climax, he explains why forecasting the climate works a lot like predicting the final league table.
Predictions for the weekend?
If we tried to predict how the Premier League table will look after this weekend’s fixtures, we would do a fairly good job. That’s because there’s only so much that can change in a single round of games. Certain teams can’t move position given the number of points they have relative to other teams. That means unfortunately for Wolves, they will stay bottom. For the teams that can move, we can have a good guess at the outcome of each match and come up with a reasonable prediction of the whole table.
Predicting the weather this weekend is much same. Just as there’s only so much the table can shift in one weekend, there are only so many things that can happen in the atmosphere given the current atmospheric circulation patterns. We can solve equations which determine the movement of air, just as we can calculate the most likely results between teams and work out their positions. Of course, there will be errors, we won’t predict every score precisely, but generally, our predictions will be close to reality.
Who’s going to win the title?
Predicting the final league table positions at the end of the season is harder. A lot can happen in a couple of months, and predicting every remaining fixture accurately becomes nearly impossible. An error here or there could have a cascading influence on final positions. Weather forecasting has the same problem; small errors compound, and predictions weeks to months ahead involve large uncertainties.
But weather prediction also has an additional challenge that football doesn’t. In the league, we know precisely where each team stands right now. In meteorology, the world is very big and accurately recording the current state of the weather worldwide is near impossible. This is what we call initial condition uncertainty. Even if you calculate every equation perfectly, if your starting picture is wrong, your future predictions will be too.
To tackle these uncertainties, forecasters take an ensemble approach; making lots of predictions based on slightly different but plausible starting outcomes. In football terms, this means running through many possible combinations of remaining results to see which final tables come up most often. Each outcome is possible, but we can’t say which will occur for sure. With a big enough ensemble, we can calculate the most likely final positions, whether in the table or in the atmosphere.
This season’s a write off, I’m focussing on next year now
How would we predict the final table for next year? A first estimate would be the average position of every team over the last 20 years. In weather forecasting, this is referred to as the climatology; the historical average of weather conditions at a place. But we can do better. Using historical averages would give extra weighting to teams which have previously performed well but are not playing well recently. Spurs, for example, would sit near the top based on past performance, but we know they are adjusting to life without key players. To improve our predictions, we would adjust our table to reflect the current form of each team.
Predicting the weather a year out follows the same logic. We know the probability of different weather conditions based on historical averages. However, slowly changing patterns, what climate scientists refer to as natural variability, affect probabilities. For example, temperatures in the Pacific ocean can influence UK weather. We can predict these patterns sometimes years in advance, giving us better information about future weather.
A turn up for the books
There are always surprises. No one predicted Leicester City winning the title in the 2015/2016 season. Models are not perfect and the real world does not always follow the rules. Similarly, individual extreme weather events are impossible to predict a year out, just as Leicester’s later success would have been impossible to foresee after they were nearly relegated in 2014/2015.
But if unusual events keep happening and predictions which used to work start being wrong all the time, that signals something has changed. For instance, prior to their takeover in 2008, Manchester City’s average Premier League position was 7th. Based on that record, we would not have predicted them to win six titles in the last decade. One or two unexpected wins can be put down to a star player or coach. But a decade of dominance isn’t luck, it’s systemic change, explained by massive investment and a world-class coach.
A lot of work has gone into understanding whether the increase in global temperature since the 1980s is natural variability or systemic change in climate. Climate scientists tackle this through models: one predicting the weather with human greenhouse gas emissions, one with only natural sources such as volcanoes. The result is clear, only the model with increasing greenhouse gases can reproduce the observed warming. Without human emissions in the model, the numbers simply don’t match what we’ve measured. From this, we can say with high confidence that the temperature increase is a result of human greenhouse gas emissions.
Unfortunately for Spurs, the link between money spent on players and Premier League position is weaker than the relationship between greenhouse gases and global temperature.