Effect of Changing Observation Time on Mean Temperature

Hourly temperature data for eight first-order stations in the United States have been used to determine the effect of observation time on mean temperature derived in the usual way from 24-hr maximum and minimum values. Results, presented in detail for exemplary stations and observation times, show that the greatest possible effect on temperature of arbitrary time changes varies with place and season between less than 1/2F and more than 3F at the stations investigated. A means of estimating both this maximum effect, and the effect of any specific time change, at an arbitrary location is presented. It is concluded that historical temperature data based on evening observations are apt to be more homogeneous than those based on morning observations. An example of the effect of typical observation time changes on a secular temperature series is presented, and stresses the value of thorough documentation of observation time by each cooperative observer.