There are different methods used in literature: For example the “threshold method” – daily pollen concentration reaches a particular threshold, “the sum method” – the accumulated daily sum reaches a particular threshold, “ per cent method” when the accumulated sum reaches one or more per cent of the annual total. The drawback of defining the season using a fixed threshold value may imply that the season does not start at all in small seasons. The per cent method cannot be used until the end of season. Thus, it depends on the objective (of your analysis) as well as on the taxa with its typical flowering and pollen release. Which taxa do you want to analyze?
A good overview is given by Jato et al. (2006) May the definition of pollen season influence aerobiological results? – Aerobiologia 22, 13—25. http://dx.doi.org/10.1007/s10453-005-9011-x.
Totally agree with Chritina on the definitions of season starts. For certain methods or definitions when using ragweed or grasses for example the start would end up very much after people are reacting to the allergen. What is the purpose of season definition. We usually use 3-4 days with counts.