Small and spatially heterogeneous areas are where climate models most struggle -- but they are also, as of 2008, home to more than half of the world's population. The challenge of making projections for urbanized regions is especially acute toward the tails of the distribution, whatever the variable of interest. My interests lie in exploring how to best capture the historical climatology of extremes in all their spatiotemporal complexity, in characterizing and categorizing the meteorological patterns that produce them, and in using this knowledge to refine projections for (the remaining part of!) the 21st century. These connections between the local and regional scales, and consequently between subdaily and seasonal events, are multifaceted, and so benefit from being approached with a variety of tools, most prominently statistical downscaling and direct representation with regional climate models.