Urban temperature bias, defined to be the difference between a shelter
temperature reading of unknown but suspected urban influence and some
appropriate rural reference temperature, is estimated through the use
of polar-orbiting satellite data. Predicted rural temperatures, based
on a method developed using sounding data, are shown to be of reasona
ble accuracy in many cases for urban bias assessments using minimum te
mperature data from selected urban regions in the United States in Jul
y 1989. Assessments of predicted urban bias were based on comparisons
with observed bias, as well as independent measures of urban heat isla
nd influence, such as population statistics and urban-rural difference
s in a vegetation index. This technique provides a means of determinin
g urban bias in regions where few if any rural reference stations are
available, or where inhomogeneities exist in land surface characterist
ics or rural station locations.