Climate Station Clustering

The climate research in the United States is focused heavily on how the climate is responding to anthropogenic forcing and the implications of a changed climate on the world going forward. A number of studies focus on continental scale climate change but also at a regional level by state or National Climate Data Center(NCDC) climate divisions. The problem with national and state level climate studies isn’t the research question, it’s how relevant is this information to the public. Aggregating climate change effects over an entire state or a nation loses much of the detail that would be useful to local climate actors such as local governments but also citizens operating businesses threatened by climate change. The NCDC climate divisions, shown in Figure 1, offer a better alternative for both the public understanding of climate change but also for climate research in general. The NCDC divisions were initially created in the early 1900’s and revised several times since then, but the divisions were based on geography and not strictly on climate variables (Guttman et al., 1996). The NCDC divisions are a superior aggregation framework over the national and statewide view, but are not entirely scientific. The western climate divisions were based more on drainage basins than statistical relationships (Bunkers et al., 1995). The best way to create climate divisions would be to analyze the millions of climate observations and define regions based on similarity. The size of the problem limits the usefulness of traditional data analysis techniques, but would be perfect for cluster analysis techniques.
           
NCDC climate divisions.