Over a quarter of adults in the United States suffer from seasonal allergies, yet the broader spatiotemporal patterns in seasonal allergy trends remain poorly resolved. This knowledge gap persists due to difficulties in quantifying allergies as symptoms are seldom severe enough to warrant hospital visits. We show that we can use machine learning to extract relevant data from Twitter posts and Google searches to examine population-level trends in seasonal allergies at high spatial and temporal resolution, validating the approach against hospital record data obtained from selected counties in California, United States. After showing that internet-derived data can be used as a proxy for aeroallergen exposures, we demonstrate the utility of our approach by mapping seasonal allergy-related online activity across the 144 most populous US counties at daily time steps over an 8-year period, highlighting the spatial and temporal dynamics in allergy trends across the continental United States.