Volcano monitoring: Addressing data quality through iterative deployment

Deploying wireless sensor networks to support geophysics presents an interesting challenge. High data-rates required by geophysical instrumentation preclude continuous data collection from even moderately-sized networks. However, geoscientists are used to working directly with complete signals, and therefore uncomfortable with in-network processing that could reduce bandwidth by reporting data products. Over five years of working with seismologists we have developed a lineage of solutions driven by their scientific goals. Three field deployments have provided valuable lessons and helped drive each successive design iteration. We began by addressing datum quality, encompassing per-sample resolution, accuracy, and time synchronization. Later deployments focused on holistic data quality, which requires considering constraints limiting full data collection in order to maximize the value of the limited data retrieved. This chapter uses our three deployments to demonstrate the benefits of iteration. The first two illustrate our work on datum quality, while the last presents a new approach to optimizing overall dataset quality.