idea 1) It's possible that some subset of your area has higher-scale / better resolution aerial photographs available. Could you find these and use them to test your model?
idea 2) What geographical area / timescale are you looking at? For certain areas/times, you can get a bit of data on phenology from dated herbarium specimens (ie pressed plant collections). So, for instance, you might select a few plant species that you think are good indicators of whatever phenology you're modelling, search for them on GBIF, consider only those specimens that match your geographical area, and look at what time of year the collections have been made. Then you could look at the first, or last, or various quantiles of phenology. It's not ideal data, but each specimen represents a very direct data point (that is, you know there was at least one individual that had flowered / leafed / whatever at that exact date and georeference), and it's available without getting your feet dirty.
idea 3) Again depending on your geographical area and what measure of phenology you're interested in, you might even be able to use non-aerial photographs, if they're dated.