Statistical characteristcs can vary throughout different levels of aggregation, such as spatial unit sizes like country compared to region. To properly analyze a phenomenon, like snowfall, different scales of the phenomenon can be studied. For the most accuracy, the "scale of the analysis must be the actual scale of the phenomenon" [Montello, 2001]. Additionally, it's important to proceed with caution on making inferences across scales, which is known as the cross-level fallacy. Particularly in the example of snowfall, starting from a country scale and disaggregating down through a hierarchy into resort-scale themselves, the statistical characteristcs from the orginal units increasingly vary when broken into smaller units. This is an example of the Modifiable Areal Unit Problem (MAUP).
Aside from being cognizant of the cross-level fallacy, MAUP, and the particular phenomenon of interest, each level of scale could provide patterns explicit to different features. This is an illustration of spatial and temporal aggregation for just snowfall. Are other weather features affected by this? What implications does this have across businesses and resort ratings? How do predictive models react to different levels of scale?