Central England’s 250-Year Temperature Record Shows a Sharp Recent Warming
CET data show little long-term change before the 20th century, followed by faster warming since the 1980s, with recent decades unusually hot.
Central England’s temperature record offers an unusually long view of regional climate change. Its daily data show a clear shift: after more than a century of fluctuations without a discernible long-term trend, temperatures began rising in the 20th century, with the steepest warming in recent decades.
The Central England Temperature series, or CET, is described as the world’s longest continuous regional temperature record. It combines thermometer readings from several stations across a broad area of central England. Monthly averages begin in 1659; daily values start on 1 January 1772; and daily highs and overnight lows are available from 1878. The Met Office provides the data here.
The early record needs more caution than the later one. Thermometry was still developing in 1659, before modern temperature scales existed. Fahrenheit dates to 1724 and Celsius to 1742, and the first estimates also draw on less data. From 1772 onward, however, CET supplies more than 250 years of daily mean temperatures. A daily mean is calculated as the average of the high and overnight low.
Removing the seasonal cycle
Daily readings vary widely, from −11.9°C (10.6°F) to 28.1°C (82.6°F). Much of the pattern comes from the ordinary seasonal cycle: temperatures rise and fall through the year. A plot of daily data since 2000 makes that repeated cycle visible, while arranging temperatures by day of the year shows it more directly.
To see the longer-term climate signal, the analysis subtracts the average temperature expected for that time of year. This produces an anomaly: the difference between a reading and the seasonal average, or climatology. A baseline period defines that climatology. Here, the baseline is 1772–1899, so each anomaly indicates how much warmer or cooler a day was than the average for the same part of the year during that pre-1900 period.
Daily anomalies still jump around, ranging from −14.93°C (−26.87°F) to 12.10°C (21.78°F). But averaging anomalies across longer periods reduces much of that day-to-day noise, making it easier to assess changes over time. The annual averages reveal an upward shift, particularly in the latest years shown; the current year in the source analysis is incomplete, so its value may change.
Decades reveal the changing pattern
Ten-year averages smooth the record further. The analysis adds error bars to represent the 95% likely range of natural variation for each period. On this measure, every decade through the 1920s includes zero within its range, consistent with temperature changes that could plausibly reflect natural variation. In the first half of the 20th century, averages rose modestly, with a change the analysis finds statistically significant. Temperatures then leveled off around the middle of the century. Since the 1980s, the pace of warming has been much faster than earlier in the record.
Thirty-year averages make the broad evolution clearer, though the choice of period boundaries affects the values slightly. The first average, covering 1770–1800, contains only 28 years of observations because CET’s daily record begins in 1772. The latest average is also incomplete, with 17 years of data. The source analysis says that shifting the boundaries would change the averages somewhat while leaving the overall pattern intact.
A running mean addresses the arbitrary boundaries by calculating an average for every possible 30-year span, advancing one year at a time. Each average is placed at the midpoint of its period. This method gives a simple view of the underlying trend, but it has limits: it can retain short-lived fluctuations, and it cannot provide values for the first or last 15 years of the record because each estimate needs a full 30-year window.
A long record, with a clear recent shift
The analysis also uses a modified LOWESS smooth, a statistical method designed to estimate a trend over a similar time scale. Its result closely matches the 30-year running means where the two cover the same years, and it is used to estimate the 95% likely range of natural variation around the smoothed signal.
Read together, the methods point to a consistent history. For roughly the first 130 years of the daily record, temperatures varied within the range attributed to natural variation, with little evidence of a sustained trend. After about 1910, CET rose beyond that range. The source identifies a possible modest decline from about 1945 to 1975, followed by a sharp rise that began in the later 20th century.
The source links this timing to the accumulation of greenhouse gases in the atmosphere since the early 20th century. It also notes that the shape of the CET trend resembles the global temperature pattern over the same period. That comparison places a regional record in wider context, while CET’s central England coverage remains distinct from a global average.
CET’s value lies in the way its length separates short-term weather swings from longer changes. The recent rise stands out across annual averages, decade-scale ranges and smoothed trends. As the current decade and year remain incomplete in the analysis, their estimates can still change; the longer-term direction shown by the record is already evident.
