Climate and weather: Trends and averages
Trends show whether a set of measured values is increasing or decreasing over time. Calculating climate trends over short periods of time can often give misleading results, since the data are affected by measurement uncertainties and by short-term variability. That’s why scientists use averages taken over longer time periods to ensure the trends calculated are more reliable.
The ease of detecting a trend depends on the amount of short-term ups and downs in the measurements – known as ‘background noise’. These fluctuations can obscure the underlying trend – the ‘signal’. High levels of noise can drown out any signal, like background chatter in an auditorium drowning out a particular person’s voice. The relative magnitude of signal and noise is known as signal-to-noise ratio. A high ratio means the signal is significantly larger than the noise, making the underlying trend of the measurements obvious. Low signal-to-noise ratio makes any trend difficult to detect. In general, the lower the signal-to-noise ratio, the longer the series of measurements needed to calculate a reliable trend.
Converting a series of measurements into averages can increase the signal-to-noise ratio, making it easier to detect any underlying trends. For example, scientists often convert annual temperature or rainfall data into a sequence of 10- or 20-year averages. This reduces the background noise, because any roughly equal fluctuations in opposite directions will tend to cancel each other out in the average. So examining a series of decadal averages can enable the calculation of more reliable trends. Temperature trends usually require a series of at least 10 and usually 20 to 30 years’ worth of measurements, while trends in rainfall often require even longer time periods because rainfall is more variable and so has a lower signal-to-noise ratio.
Climate patterns are far from uniform across the globe and even neighbouring regions can show stark differences. Scientists often convert local climate conditions into averages over larger regions to help them understand the underlying patterns. Smaller regions can appear to contradict larger spatial patterns. For example, the Gulf Stream keeps UK temperatures above average for its latitude, but the large-scale pattern is one of decreasing temperature from the Equator to the Arctic. If climate changes, the trends over time are also far from uniform across the globe. It’s common for some regions to change more rapidly than others and for some areas to experience opposite trends to the global average.
In order to calculate global average climate conditions, scientists sift through data from measuring stations worldwide. There are sometimes many more measurements covering one area than another, in which case scientists adjust the weight given to each measurement, since over-representing a particular area could distort the average. Some studies estimate that trends in global average surface air temperature can be reliably calculated using measurements from just 1000 different locations, provided they are distributed widely enough around the globe. However, scientists tend to use as many different measurements as possible – records of global surface temperature currently include data from more than 4000 weather stations.