Assessment of the uncertainties in temperature change in China during the last century

Abstract

We have used the China Homogenized Historic Temperature dataset and some long-term station series of the neighbor countries from CRUTEM3, a 5°×5° gridded dataset of monthly mean temperature since 1900, to provide a 107-year record of surface temperature trends and variability. We derived a comprehensive set of uncertainty estimates to accompany the data: measurement and sampling errors, uncertainties in temperature bias estimates, and uncertainties arising from limited observational coverage on large-scale averages have all been estimated. We reanalysed the temperature changes during the period of record. The best estimates of trends for 1900–2006 with uncertainties at 95% confidence range are about 0.09±0.017°C/decade for the year as a whole, and 0.14±0.021°C/decade, 0.11±0.021°C/decade, 0.04±0.017°C/decade, and 0.07±0.017°C/decade for winter, spring, summer and autumn respectively. For 1954–2006, the trends for annual, winter, spring, summer and autumn are: 0.26±0.032°C/decade, 0.35±0.046°C/decade, 0.25±0.051°C/decade, 0.16±0.037°C/decade and 0.22±0.055°C/decade. Winter saw the most significant warming trend in both 1900–2006 and 1954–2006, while during the most recent period (the satellite era, 1979–2006), all the seasons show similar warming trends: 0.45±0.13°C/decade, 0.51±0.11°C/decade, 0.52±0.16°C/decade, 0.37±0.10°C/decade and 0.50±0.16°C/decade for annual, winter, spring, summer and autumn. Trends arising from urbanization have been evaluated as less than 5% of the total warming trend for 1951–2001, so this bias was not removed.