Mapping temperature using a Bayesian statistical method and a high accuracy surface modelling method in the Beijing–Tianjin–Hebei region, China

Abstract

A more accurate Bayesian statistical technique together with a high accuracy surface modelling method (HASM) was used to improve the accuracy of temperature fields in the Beijing–Tianjin–Hebei (BTH) region, China. Bayesian statistical inference theory was first applied to fill missing daily values for meteorological stations on an annual scale. A mixed interpolator was then employed to simulate the annual mean temperature and was compared with other methods. The annual mean temperature was produced and the inter-decade change was investigated. The results show that the values filled by Bayesian statistical inference theory agree well with actual observations. A comparison with other interpolators shows that combining the Bayesian estimation method with the HASM gives better results than those of the other methods tested in this study. The construction of the annual mean temperature for 10 years shows the change interval and the spatial distribution of temperature in the BTH region. The annual temperature over 10 years changed from less than −1 °C in the northwest to more than 14 °C in the southeast. From 1981–1990 (T1) to 1991–2000 (T2) and T2 to 2001–2010 (T3) the temperature change was spatially stationary and varied mainly from −0.05 °C to 0.05 °C, except for some areas such as the southwest of Tangshan city and the south of Tianjin city where land cover and land use has exhibited large decadal variation during the past 30 years.