I stumbled upon the "Project Weather" group on Flickr a while back, and started to wonder if the group could also be considered a relevant data source? For example, searching thunderstorms in the group produces quite interesting results - all of them geotagged (at least manually).

I think both the people submitting the photos and flickr staff selecting and curating the results are probably more motivated by esthetics than research, but how much does this kind of selection bias matter?