The models from ppmsuite agree a lot. Best agreement: - gaussian and curve (expected since they are practically the same) - sppm and gaussian, ecc) - towards the end of the similarity there is drpm with all the other ones (makes sense, they were from different packages, and drpm was the only one including time)
The summer period is the most indecisive/dubious in terms of comparison/agreement. Maybe this is because the trend in summer is very flat, so almost all stations behave in the same way and therefore there is no clear/distinct clustering definition.
In fact, noticebly, the drpm is the one disagreeing with the others (which instead also during summer tend to behave in the same way for the clustering generation), since maybe accounting time leads him to a more precise clusetering. Or anyway time is relevant so just drpm accounting for leads to different clusters with respect to models which dont include time.
Almost all models create a stratification. The planar regions show in general an high value of pm10, and are indeed frequently clustered together (like milan area or milan + mantova). While moving upwards towards the alps we find less and less polluted clusters.
There seems to be some """outliers""" (are just noticeble stations, fully explainable but emerging) in the south west area wihch are in all models clustered together and separately from the surroinding cluster of milan. Indeed they have lower pm10 cooncentrations and high levels of LA_hvi (c'hanno molti alberi).
Going vertically, we find several recurrent patterns of clusters: the alps, the pre-alps,