The Use of Regression Equations to Increase the Usefulness of Historical Temperature Data in On-line Pest Management 1

Publisher Website
Google Scholar
Abstract
The density of stations in on-line weather monitoring networks for pest management is likely to be rather low because of the cost of maintaining them. Multiple and simple linear regression techniques provide a means of estimating missing data, and of estimating temperatures at the much denser “a” network stations of the National Weather Service. In general, the improvement in the multiple correlation coefficient by using more than 2 stations as predictors is marginal, and when the resulting temperatures are used to compute degree-days, one predictor station is nearly as good as 14. That is true whether one predictive equation is used for the season or if new equations for consecutive 15-day periods in the season are used. Computer-generated maps provide a convenient means for estimating degree-day accumulation at locations between weather stations. By using regression equations to estimate temperatures for stations that are not on-line, it is possible to increase the resolution of the accumulated degree-day maps.

This publication has 1 reference indexed in Scilit: