Regionalization of hydroclimate variables in the contiguous United States

We apply a hierarchical clustering algorithm to the Parameter-elevation Relationships on Independent Slopes Model (PRISM) database. The method employs linkage clustering while forcing spatial contiguity. We apply it to the lower-48 United States, deriving regions that are based on temperature and precipitation averages and anomalies, as well as statistical parameters underlying several drought and intense precipitation measures. Resulting regions make intuitive sense from the perspective of driving influences on temperature and precipitation averages and anomalies, and are compatible with results from another empirically derived clustering scheme. Regions selected for individual variables show high similarity across different time frames. There is slightly less similarity when comparing regions created for different monthly or daily hydroclimate variables, and relatively low similarity between monthly vs. daily measures. It is unlikely that any one regionalization solution could summarize hydroclimate extremes given the wide range of variables used to describe them, but geographically sensitive datasets like PRISM and flexible algorithms provide useful methods for regionalization that can aid in drought monitoring and forecasting, and with impacts and planning associated with heavy precipitation.