Sun. Oct 2nd, 2022
An interesting way of looking at the world: the darker the red, the closer the correlation between the local temperature and the global mean.  Blue areas tend to prefer opposite temperatures.
enlarge An interesting way of looking at the world: the darker the red, the closer the correlation between the local temperature and the global mean. Blue areas tend to prefer opposite temperatures.

Meteorologists use weather forecasting models to make accurate predictions of the weather conditions for the coming days. In contrast, climate scientists use global climate models to predict the effects of climate-changing greenhouse gas emissions in the coming decades. In between these two activities is an interesting task that has proved more difficult than both: predicting global temperatures in a few years’ time.

A new study by Patrick Brown and Ken Caldeira tries a new approach to this challenge with nothing more than statistical analysis of the last two year temperatures.

The annual mean surface temperature for the world varies a bit from year to year, even if a long-term warming trend is apparent. It is those fluctuations from year to year that are difficult to predict. They depend on variable regional weather patterns, most notably the El Niño Southern Oscillation. This seesaw pattern of warm surface water along the equatorial Pacific Ocean is significant enough to cause the planet’s average surface temperature to rise and fall. It also affects weather patterns in many places around the world.

And so are other ways in which large-scale weather patterns vary. Years in which the planet sets a new temperature record are, of course, also warm on a local scale in most places. So the ability to provide advance warning is more useful than just accounting for global temperature.

This is clearly something climate scientists have been toying with for a long time. One way to do this is to feed recent temperature data into a climate model and then simulate it a year or two ahead. That has some advantages, but the fluctuations of El Niño/La Niña are stubbornly difficult to nail in simulation and these models require supercomputer time.

Another less labor-intensive option is to make a purely statistical prediction based on previous data. This is often based on metrics of variability, such as the El Niño Southern Oscillation Index, which distills sea surface temperature patterns into a number that represents the states of La Niña, neutral, or El Niño. But while useful, that power discarding data that could provide a better prediction.

A new method

So to try something different, the researchers constructed a method that simply records temperature data for each cell in a grid around the world. For all data before the year 2000, their method analyzes the full temperature pattern over a two-year period and compares it to the average global temperature in next year. The end result is a complex mathematical correlation that can be used to predict future temperatures on Earth.

The method has been tested in several ways. It was repeatedly recalculated on the basis of each year before 2000 except one† Then the method was tested by predicting that missing year – what is known as ‘leave-one-out’ validation. It was also used to predict global temperatures for every year after 2000, which had not been used to calculate correlations. In any case, the method performed reasonably well, beating simple assumptions of a sustained trend, as well as climate model simulations.

If we take a closer look at the correlation patterns, the method does nothing surprising. It picks up the trend of global warming temperatures and also finds predictive power in the areas of the ocean that vary with things like the El Niño Southern Oscillation. It shows that an El Niño (warm water extending to the eastern side of the equatorial Pacific) is associated with warmer global temperatures the following year. The cooler waters of a La Niña, on the other hand, are usually associated with a warmer global temperature two years afterwards – due to the back and forth nature of this phenomenon.

In addition to the study, the researchers posted up-to-date forecasts online, with new data every month to predict the next three years. (You can also see how the method would have predicted each year before.) It currently predicts that 2020 has a 92 percent chance of becoming the warmest year on record. The year is currently halfway through, but that’s comparable to Berkeley Earth’s current forecast based on temperature so far. On the other hand, NOAA’s latest forecast — released Thursday — put those odds at 49 percent.

Hereby the latest version of the forecasts for 2020-2022.
enlarge Hereby the latest version of the forecasts for 2020-2022.

Of course, this statistical approach cannot predict unforeseen events. Volcanic eruptions (or global pandemics) can cause sudden ripples that have little to do with the previous year’s temperatures. Take, for example, the method’s updated 2020 forecast. In the published study, it was presented based on data up to the end of 2019. There, the central estimate would have registered it as the third-warmest year on record, although the error bars certainly include the current forecast.

The first half of 2020 was quite warm, pushing the current forecast upward. That could just be a matter of the method missing the mark a bit, but it could be that the economic effects of COVID-19 have led to a reduction in aerosol pollution and thus a little extra heat. The possibility has not yet been explored in detail.

Regardless, this study adds another independent method that could lead to more reliable near-term global temperature predictions. That would mean more of a warning when a record year is on deck, including the first year to surpass a milestone like 1.5°C warming.

Earth and Space Science, 2020. DOI: 10.1029/2020EA001116 (About DOIs).

By akfire1

Leave a Reply

Your email address will not be published.