The United States Historical Climatology Network (HCN) serial temperature d
ataset is comprised of 1211 high-quality, long-term climate observing stati
ons. The HCN dataset is available in several versions, one of which include
s population-based temperature modifications to adjust urban temperatures f
or the "heat-island" effect. Unfortunately, the decennial population metada
ta file is not complete as missing values are present for 17.6% of the 12 2
10 population values associated with the 1221 individual stations during th
e 1900-90 interval. Retrospective grid-based populations; within a fixed di
stance of an HCN station, were estimated through the use of a gridded popul
ation density dataset and historically available U.S. Census county data. T
he grid-based populations for the HCN stations provide values derived from
a consistent methodology compared to the current HCN populations that can v
ary as definitions of the area associated with a city change over time. The
use of grid-based populations may minimally be appropriate to augment popu
lations for HCN climate stations that lack any population data, and are rec
ommended when consistent and complete population data are required. The rec
ommended urban temperature adjustments based on the HCN and grid-based meth
ods of estimating station population can be significantly different for ind
ividual stations within the HCN dataset.