A Study of Temperature Trends Using Rural Sites Identified from MODIS Classifications

Does urban heating effect the estimates of global average land surface temperature? We study this possible source of bias by applying an urban-rural classification based on MODIS satellite data to the Berkeley Earth temperature dataset compilation of 39,028 sites from 10 different publicly available sources. We identify a a rural subset of 16,132 sites chosen to be distant from MODIS-identified urban areas. The distribution of linear temperature trends for this subset of sites is compared to the distribution for all sites. While the trend distributions are broad, with one-third of the stations in the US and worldwide having a negative trend, both distributions show significant warming. For our primary analysis, we estimate time series of the Earth's average land temperature using the Berkeley Earth methodology applied to the full dataset and the rural subset. We observe the opposite of an urban heating effect over the period 1950 to 2010, with a slope of -0.19 ± 0.19 °C / century (95% confidence), a comparably small effect as the estimates published by previous groups. Given the small size of the effect, and its negative sign, we conclude that urban warming does not substantially affect estimates of recent global warming -- confirming the key conclusion of the previous groups' work.