Tagging a small area (e.g. Montlake) from a large vector data source (e.g. the Seattle 2ft contours) entails spending almost as long reading the vector data in as doing the actual processing. Is there a way to read only the relevant subset of the vector data source?