Airborne pollen can trigger allergic reactions and threaten public health. Better predictions of airborne pollen are needed, but to create them we must understand when plants are flowering. Satellite imagery could be useful for this, but it usually can't detect flowers directly. It can, however, detect leaf out of trees. Here, we estimate leaf out time for hundreds of thousands of trees in New York City using satellite imagery, and then use citizen science observations of the difference between tree leaf out and flowering time to predict when those trees will flower. This method worked well for many species and allowed us to predict the timing of flowering for those trees over space and time. There was substantial variation in flowering time among species and over space that we previously wouldn't have been able to understand. We expect that this method of estimating flowering time will be useful for making models of airborne pollen in New York City and beyond.