Predicting the onset of disease before it hits the orchard is an important and challenging task. Plant pathologists use scientific models that take into account environmental factors such as temperature and leaf wetness to predict when conditions are ideal for infections to occur, thereby allowing growers to take preventive action. Currently, the models rely for the most part on data from weather stations that may not necessarily be located in or close to the orchard. We are working toward improving prediction accuracy by increasing data collection density both in time and space. We have installed a 10-node wireless sensor network consisting of ten weather stations in an orchard to measure the variability of temperature and leaf wetness inputs across locations and to compare them to data from a single weather station located in the periphery of the orchard. The nodes communicate wirelessly with a remote base station, and users can access the data via a web interface. We will verify how the dense sensor network dataset compares to the single station when predicting the onset of apple scab using the Modified Mills Table model. The lessons learned should allow us to determine whether there should be any modifications to the way scientists approach disease prediction.