A demonstration that large-scale warming is not urban

On the premise that urban heat islands are strongest in calm conditions but are largely absent in windy weather, daily minimum and maximum air temperatures for the period 1950-2000 at a worldwide selection of land stations are analyzed separately for windy and calm conditions, and the global and regional trends are compared. The trends in temperature are almost unaffected by this subsampling, indicating that urban development and other local or instrumental influences have contributed little overall to the observed warming trends. The trends of temperature averaged over the selected land stations worldwide are in close agreement with published trends based on much more complete networks, indicating that the smaller selection used here is sufficient for reliable sampling of global trends as well as interannual variations. A small tendency for windy days to have warmed more than other days in winter over Eurasia is the opposite of that expected from urbanization and is likely to be a consequence of atmospheric circulation changes.