Errors in Estimating Monthly Degree Day Normals by the Fast Method

Abstract

The non-linear part of the relationship between monthly degree days and temperature was studied by use of 23 years of daily temperature data at 64 stations representing all major climatic zones in the co-terminous U.S. The data reveal that errors in estimating monthly degree days by the fast method range from a few percent to more than 300%, when the differences between the base and the mean temperature range, respectively, from + 8 to ? 2°C (+ 14 to ? 4°F). The errors can be reduced to less than 10% if the fast method is used together with an additional parameters, the estimated standard deviation of the daily mean temperature for the period of interest, and an appropriate model for the temperature distribution. The exact expression, valid for any model, is given for the relationship between degree day and mean temperature normals.