Improved methods for estimating effective accumulated temperature for pest forecasting

Abstract

　The optimal timing of pest control is forecasted via effective accumulated temperature. We focused on the daily increment of it (D_y,i ; y and i indicate the year and day, respectively). D_y,i is usually estimated using the hourly temperature. The observed hourly temperature data are only available for the time period before the publication date (when the pest control information is released). Normals (historical data from a 30-year period) of hourly temperature have been used for the time period after the publication date, which may lead to errors in estimating D_y,i. We examined the errors caused by the use of normals and developed new methods for estimating D_y,i for the time period after the publication date: the normals of D_y,i. Simulations for the past 30 years showed that pest forecasting using the normals of D_y,i resulted in smaller errors than the existing method because it reflected diurnal temperature variations. We proposed another method for cases where high temperatures are predicted in the time period after the publication date and showed an example in which this method had better accuracy than the existing method.