Navigating oceans of data

Some science domains have the advantage that the bulk of the data comes from a single source instrument, such as a telescope or particle collider. More commonly, big data implies a big variety of data sources. For example, the Center for Coastal Margin Observation and Prediction (CMOP) has multiple kinds of sensors (salinity, temperature, pH, dissolved oxygen, chlorophyll A & B) on diverse platforms (fixed station, buoy, ship, underwater robot) coming in at different rates over various spatial scales and provided at several quality levels (raw, preliminary, curated). In addition, there are physical samples analyzed in the lab for biochemical and genetic properties, and simulation models for estuaries and near-ocean fluid dynamics and biogeochemical processes. Few people know the entire range of data holdings, much less their structures and how to access them. We present a variety of approaches CMOP has followed to help operational, science and resource managers locate, view and analyze data, including the Data Explorer, Data Near Here, and topical "watch pages." From these examples, and user experiences with them, we draw lessons about supporting users of collaborative "science observatories" and remaining challenges.