We can then draw a more general conclusion from our results: that the spatial and temporal scales for prediction must be chosen carefully, not only to match the scales of disease spread [54] [55] [56] , but also with respect to the scales of the weather events that might affect the spread.