In this talk I will describe the result of using a variant of Principal Component Analysis called Archetypal Analysis (Cutler and Breiman, 1994) to split the spatial and temporal parts of a time series in 56 dimensions.  The spatial part are characteristic patterns of outbreak across the counties of the state of Montana, and the temporal part describes which pattern appears at what point of time.   A data set of seasonal flu case counts over 8 years is compared to the data set of COVID 19 over the three different outbreaks we have experienced in the last two years.  The mutual information between different counties during the outbreaks is used to initially pull out the most “influential” counties and reduce the dimension of the data set analyzed by archetypes.