Abstract
A method to estimate missing daily maximum and minimum temperatures is presented. Temperature estimates are based on departures from daily temperature normals at the three closest stations with similar observation times. Although applied to Cooperative Observer Network stations in the northeastern United States, the approach can be used with any network of stations possessing an adequate station density and period of record.
Generally, 75% of the estimates for both daily maximum and minimum temperature are within 1.7°C of the observed value. Median absolute differences between estimated and observed minimum temperatures, however, tend to be greater than those associated with maximum temperatures. For minimum temperatures, median absolute differences are approximately 1.0°C, whereas for maximum temperatures these differences are near 0.5°C. The accuracy of the estimates is independent of observation time, geographic location, and observed temperature but is influenced somewhat by station density.
