Urbanization Effect in Regional Temperature Series Based on a Remote Sensing Classification Scheme of Stations

Quantifying the urbanization effect on station and regional surface air temperature (SAT) trends is a prerequisite for monitoring and detecting long-term climate change. Based on the data set of satellite visible spectral remote sensing, a new method is developed to determine the urbanization level around observational sites on varied spatial scales and to classify the sites into different categories of stations (U1, U2, horizontal ellipsis , U6) with U1 the least and U6 the largest affected by urbanization. Urbanization effect on SAT anomaly series of urban and national stations are then evaluated for the periods of 1980-2015 and 1960-2015. Results show that the percentage of built-up area in different circumferences of the observational sites can be considered as a good indicator of comprehensive urbanization level of station and can be used to classify stations and to determine reference stations; the largest increase in annual mean SAT (T-mean) during 1980-2015 occurred at U6 stations, and U1 stations registered the weakest annual mean warming. The urbanization level is significantly positively correlated to the linear trends of annual mean T-mean and minimum SAT (T-min) and significantly negatively correlated to the diurnal temperature range (DTR) change. The data sets of the national reference climate station network and basic meteorological station network show large urbanization effect and contribution, with the annual mean urbanization contributions reaching 28.7% and 25.8% for the periods 1960-2015 and 1980-2015, respectively. For all the national stations (2,286 in total), the urbanization contributions are 17.1% and 14.6% for the two same periods, respectively.