We extend a previously developed method that quantifies the sensitivity of the exponential epidemic growth rate, r, to weather changes, through a pathogen's life cycle components (basic reproduction number, latent period, and mean and standard deviation of the spore production curve). Here a method is developed to study the elasticities of the system and subsequently the model is linked to observed weather patterns. This enables a direct comparison between the effects of different weather variables (temperature, surface wetness duration, and light quantity) under realistic weather scenarios. The three sites studied represent areas within the United Kingdom with contrasting climates. Yellow rust, caused by Puccinia striiformis, on winter wheat is studied as a key application. Our results show that temperature and more importantly changes in temperature through their effect on pathogen reproduction have the largest effect on r. The long latent period at low winter temperatures is not a key component in the epidemic development, which is contrary to general beliefs. The results combined with long term average yellow rust severity patterns show that it is winter survival and not summer survival that controls the eventual disease severity. The results also show that within the current United Kingdom spraying regime on wheat crops against yellow rust, the first spray should mainly affect the basic reproduction number, i.e., should be a protectant spray, whereas the second spray should also affect the latent period, i.e., should also have curative action.