Regional differences in surface air temperature changing patterns from 1960 to 2016 of China

Abstract

A thorough understanding of regional differences in change patterns of surface air temperature (SAT) at various spatial scales can help people cope well with global warming. In this study, a SAT change pattern recognition method was firstly established based on k-mean++ algorithm. By creating a clustering effect evaluation index (DBWk) to select the optimal cluster number k, the pattern of SAT change in 1960–2016 of China was recognized at national and regional scales. Results showed that China’s SAT change patterns were grouped into 3 clusters, namely, Clusters I, II and III, at the national scale. These clusters were further divided into 3, 7, and 4 subclusters at the regional scale, respectively. The SAT change in Cluster I was intense, with a relatively cold period (1960–1987) and a relatively warm period (1988–2016). The SAT of Cluster II decreased slightly in the first phase (1960–1983), minimally increased in the third phase (1999–2016), but rose strongly in the second phase (1984–1998). The linear trend (LT) of SAT increase of Cluster III was high and statistically significant, especially in 1983–2016. The analysis of the SAT change pattern of subclusters showed that the SAT fluctuations of the Altai Mountains and Junggar Basin were the strongest. The Northern Qinghai–Tibet Plateau had the highest warming rate, and the LT of warming was statistically the most significant in China.