Analysis of surface air temperature warming rate of China in the last 50 years (1962–2011) using k-means clustering

Abstract

Using observed surface temperature data from about 570 meteorological stations in China, the linear temperature trend over the last 50 years (1962–2011) is estimated. Average warming rates at all stations are 0.284 ± 0.142, 0.200 ± 0.163, and 0.392 ± 0.193 °C/10 a for annual, June–July–August (JJA), and December–January–February (DJF) mean temperatures, respectively. The warming rates (°C/a) have a regressed slope of 0.001 with latitude, and their relationship with latitude using data from stations below 1,000 m altitude is estimated at 0.009 °C/(10 a · degree). A k-means clustering is constructed using correlation coefficients to quantify the closeness. The clustering schemes with k = 5 and k = 10 are successfully implemented for annual, DJF, and JJA mean temperatures. The k = 10 scheme showed greater ability to distinguish different locations than the k = 5 clustering. In both schemes, the mean estimated warming rates are averaged over stations in each cluster. The clusters in DJF are much different than those for the entire year and for JJA. The average warming rate in each cluster is estimated and compared. Using stations in the cluster of high-elevation southeast Tibetan Plateau, the relationship between warming rates and altitude is estimated at 1.0 × 10−2 °C/(10 a · 100 m).